Datasets:

mj33

/

SimCoPilot

like 0

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '']

[' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix']

['', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 100 is imported at line 3 and has a Long-Range dependency. Variable 'scale_factor' used at line 100 is defined at line 94 and has a Short-Range dependency. Variable 'costheta' used at line 100 is defined at line 97 and has a Short-Range dependency. Variable 'sintheta' used at line 100 is defined at line 98 and has a Short-Range dependency. Variable 'dx' used at line 100 is defined at line 94 and has a Short-Range dependency. Variable 'scale_factor' used at line 101 is defined at line 94 and has a Short-Range dependency. Variable 'sintheta' used at line 101 is defined at line 98 and has a Short-Range dependency. Variable 'costheta' used at line 101 is defined at line 97 and has a Short-Range dependency. Variable 'dy' used at line 101 is defined at line 94 and has a Short-Range dependency. Variable 'similarity_matrix' used at line 103 is defined at line 100 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 9}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):']

[' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result']

['', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 106 is imported at line 3 and has a Long-Range dependency. Variable 'scale' used at line 106 is defined at line 105 and has a Short-Range dependency. Library 'np' used at line 107 is imported at line 3 and has a Long-Range dependency. Variable 'ax' used at line 107 is defined at line 105 and has a Short-Range dependency. Variable 'ay' used at line 107 is defined at line 105 and has a Short-Range dependency. Library 'np' used at line 108 is imported at line 3 and has a Long-Range dependency. Library 'np' used at line 109 is imported at line 3 and has a Long-Range dependency. Function 'translation' used at line 109 is defined at line 82 and has a Medium-Range dependency. Variable 'x' used at line 109 is defined at line 105 and has a Short-Range dependency. Variable 'y' used at line 109 is defined at line 105 and has a Short-Range dependency. Function 'rotation' used at line 109 is defined at line 86 and has a Medium-Range dependency. Variable 'angle' used at line 109 is defined at line 105 and has a Short-Range dependency. Library 'np' used at line 110 is imported at line 3 and has a Long-Range dependency. Variable 'scaling' used at line 110 is defined at line 106 and has a Short-Range dependency. Variable 'result' used at line 110 is defined at line 109 and has a Short-Range dependency. Library 'np' used at line 111 is imported at line 3 and has a Long-Range dependency. Variable 'shear' used at line 111 is defined at line 107 and has a Short-Range dependency. Variable 'result' used at line 111 is defined at line 110 and has a Short-Range dependency. Variable 'result' used at line 112 is defined at line 111 and has a Short-Range dependency.

{}

{'Library Long-Range': 6, 'Variable Short-Range': 11, 'Function Medium-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):']

[' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])']

[' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 106 is imported at line 3 and has a Long-Range dependency. Variable 'scale' used at line 106 is defined at line 105 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])']

[' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])']

[' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 107 is imported at line 3 and has a Long-Range dependency. Variable 'ax' used at line 107 is defined at line 105 and has a Short-Range dependency. Variable 'ay' used at line 107 is defined at line 105 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])']

[' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result']

['', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 109 is imported at line 3 and has a Long-Range dependency. Function 'translation' used at line 109 is defined at line 82 and has a Medium-Range dependency. Variable 'x' used at line 109 is defined at line 105 and has a Short-Range dependency. Variable 'y' used at line 109 is defined at line 105 and has a Short-Range dependency. Function 'rotation' used at line 109 is defined at line 86 and has a Medium-Range dependency. Variable 'angle' used at line 109 is defined at line 105 and has a Short-Range dependency. Library 'np' used at line 110 is imported at line 3 and has a Long-Range dependency. Variable 'scaling' used at line 110 is defined at line 106 and has a Short-Range dependency. Variable 'result' used at line 110 is defined at line 109 and has a Short-Range dependency. Library 'np' used at line 111 is imported at line 3 and has a Long-Range dependency. Variable 'shear' used at line 111 is defined at line 107 and has a Short-Range dependency. Variable 'result' used at line 111 is defined at line 110 and has a Short-Range dependency. Variable 'result' used at line 112 is defined at line 111 and has a Short-Range dependency.

{}

{'Library Long-Range': 3, 'Function Medium-Range': 2, 'Variable Short-Range': 8}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))']

[' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result']

['', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 110 is imported at line 3 and has a Long-Range dependency. Variable 'scaling' used at line 110 is defined at line 106 and has a Short-Range dependency. Variable 'result' used at line 110 is defined at line 109 and has a Short-Range dependency. Library 'np' used at line 111 is imported at line 3 and has a Long-Range dependency. Variable 'shear' used at line 111 is defined at line 107 and has a Short-Range dependency. Variable 'result' used at line 111 is defined at line 110 and has a Short-Range dependency. Variable 'result' used at line 112 is defined at line 111 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 5}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1']

[' y1 = int(y)', ' y2 = y1 + 1']

['', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'y' used at line 117 is defined at line 114 and has a Short-Range dependency. Variable 'y1' used at line 118 is defined at line 117 and has a Short-Range dependency.

{}

{'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]']

[' f21 = image[y2][x1]', ' f22 = image[y2][x2]']

['', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Else Reasoning', 'usage_line': 125}, {'reason_category': 'Else Reasoning', 'usage_line': 126}]

Variable 'image' used at line 125 is defined at line 114 and has a Medium-Range dependency. Variable 'y2' used at line 125 is defined at line 118 and has a Short-Range dependency. Variable 'x1' used at line 125 is defined at line 115 and has a Short-Range dependency. Variable 'image' used at line 126 is defined at line 114 and has a Medium-Range dependency. Variable 'y2' used at line 126 is defined at line 118 and has a Short-Range dependency. Variable 'x2' used at line 126 is defined at line 116 and has a Short-Range dependency.

{'Else Reasoning': 2}

{'Variable Medium-Range': 2, 'Variable Short-Range': 4}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)']

[' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)']

['', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Else Reasoning', 'usage_line': 129}, {'reason_category': 'Else Reasoning', 'usage_line': 130}, {'reason_category': 'Else Reasoning', 'usage_line': 131}]

Variable 'x' used at line 129 is defined at line 114 and has a Medium-Range dependency. Variable 'x1' used at line 129 is defined at line 115 and has a Medium-Range dependency. Variable 'y2' used at line 129 is defined at line 118 and has a Medium-Range dependency. Variable 'y' used at line 129 is defined at line 114 and has a Medium-Range dependency. Variable 'x2' used at line 130 is defined at line 116 and has a Medium-Range dependency. Variable 'x' used at line 130 is defined at line 114 and has a Medium-Range dependency. Variable 'y' used at line 130 is defined at line 114 and has a Medium-Range dependency. Variable 'y1' used at line 130 is defined at line 117 and has a Medium-Range dependency. Variable 'x' used at line 131 is defined at line 114 and has a Medium-Range dependency. Variable 'x1' used at line 131 is defined at line 115 and has a Medium-Range dependency. Variable 'y' used at line 131 is defined at line 114 and has a Medium-Range dependency. Variable 'y1' used at line 131 is defined at line 117 and has a Medium-Range dependency.

{'Else Reasoning': 3}

{'Variable Medium-Range': 12}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '']

[' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)']

['', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'w1' used at line 133 is defined at line 128 and has a Short-Range dependency. Variable 'f11' used at line 133 is defined at line 123 and has a Short-Range dependency. Variable 'w2' used at line 133 is defined at line 129 and has a Short-Range dependency. Variable 'f12' used at line 133 is defined at line 124 and has a Short-Range dependency. Variable 'w3' used at line 133 is defined at line 130 and has a Short-Range dependency. Variable 'f21' used at line 133 is defined at line 125 and has a Short-Range dependency. Variable 'w4' used at line 133 is defined at line 131 and has a Short-Range dependency. Variable 'f22' used at line 133 is defined at line 126 and has a Short-Range dependency.

{}

{'Variable Short-Range': 8}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):']

[' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output']

['', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Define Stop Criteria', 'usage_line': 141}, {'reason_category': 'Loop Body', 'usage_line': 142}, {'reason_category': 'Define Stop Criteria', 'usage_line': 142}, {'reason_category': 'Loop Body', 'usage_line': 143}, {'reason_category': 'Loop Body', 'usage_line': 144}, {'reason_category': 'Loop Body', 'usage_line': 145}, {'reason_category': 'Loop Body', 'usage_line': 146}]

Variable 'I' used at line 136 is defined at line 135 and has a Short-Range dependency. Library 'np' used at line 137 is imported at line 3 and has a Long-Range dependency. Variable 'rows' used at line 137 is defined at line 136 and has a Short-Range dependency. Variable 'cols' used at line 137 is defined at line 136 and has a Short-Range dependency. Variable 'cols' used at line 138 is defined at line 136 and has a Short-Range dependency. Variable 'rows' used at line 138 is defined at line 136 and has a Short-Range dependency. Library 'np' used at line 139 is imported at line 3 and has a Long-Range dependency. Variable 'T' used at line 139 is defined at line 135 and has a Short-Range dependency. Variable 'rows' used at line 141 is defined at line 136 and has a Short-Range dependency. Variable 'cols' used at line 142 is defined at line 136 and has a Short-Range dependency. Library 'np' used at line 143 is imported at line 3 and has a Long-Range dependency. Variable 'j' used at line 143 is part of a Loop defined at line 142 and has a Short-Range dependency. Variable 'center' used at line 143 is defined at line 138 and has a Short-Range dependency. Variable 'i' used at line 143 is part of a Loop defined at line 141 and has a Short-Range dependency. Library 'np' used at line 144 is imported at line 3 and has a Long-Range dependency. Variable 'T_invert' used at line 144 is defined at line 139 and has a Short-Range dependency. Variable 'shift_center' used at line 144 is defined at line 143 and has a Short-Range dependency. Variable 'coordinates' used at line 145 is defined at line 144 and has a Short-Range dependency. Variable 'center' used at line 145 is defined at line 138 and has a Short-Range dependency. Variable 'output' used at line 146 is defined at line 137 and has a Short-Range dependency. Variable 'i' used at line 146 is part of a Loop defined at line 141 and has a Short-Range dependency. Variable 'j' used at line 146 is part of a Loop defined at line 142 and has a Short-Range dependency. Function 'bilinear_interpolation' used at line 146 is defined at line 114 and has a Long-Range dependency. Variable 'I' used at line 146 is defined at line 135 and has a Medium-Range dependency. Variable 'x' used at line 146 is defined at line 145 and has a Short-Range dependency. Variable 'y' used at line 146 is defined at line 145 and has a Short-Range dependency. Library 'np' used at line 147 is imported at line 3 and has a Long-Range dependency. Variable 'output' used at line 147 is defined at line 137 and has a Short-Range dependency. Variable 'output' used at line 148 is defined at line 147 and has a Short-Range dependency.

{'Define Stop Criteria': 2, 'Loop Body': 5}

{'Variable Short-Range': 18, 'Library Long-Range': 5, 'Variable Loop Short-Range': 4, 'Function Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))']

[' center = (cols/2, rows/2)']

[' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'cols' used at line 138 is defined at line 136 and has a Short-Range dependency. Variable 'rows' used at line 138 is defined at line 136 and has a Short-Range dependency.

{}

{'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)']

[' T_invert = np.linalg.inv(T)']

['', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Library 'np' used at line 139 is imported at line 3 and has a Long-Range dependency. Variable 'T' used at line 139 is defined at line 135 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):']

[' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output']

['', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 143}, {'reason_category': 'Loop Body', 'usage_line': 144}, {'reason_category': 'Loop Body', 'usage_line': 145}, {'reason_category': 'Loop Body', 'usage_line': 146}]

Library 'np' used at line 143 is imported at line 3 and has a Long-Range dependency. Variable 'j' used at line 143 is part of a Loop defined at line 142 and has a Short-Range dependency. Variable 'center' used at line 143 is defined at line 138 and has a Short-Range dependency. Variable 'i' used at line 143 is part of a Loop defined at line 141 and has a Short-Range dependency. Library 'np' used at line 144 is imported at line 3 and has a Long-Range dependency. Variable 'T_invert' used at line 144 is defined at line 139 and has a Short-Range dependency. Variable 'shift_center' used at line 144 is defined at line 143 and has a Short-Range dependency. Variable 'coordinates' used at line 145 is defined at line 144 and has a Short-Range dependency. Variable 'center' used at line 145 is defined at line 138 and has a Short-Range dependency. Variable 'output' used at line 146 is defined at line 137 and has a Short-Range dependency. Variable 'i' used at line 146 is part of a Loop defined at line 141 and has a Short-Range dependency. Variable 'j' used at line 146 is part of a Loop defined at line 142 and has a Short-Range dependency. Function 'bilinear_interpolation' used at line 146 is defined at line 114 and has a Long-Range dependency. Variable 'I' used at line 146 is defined at line 135 and has a Medium-Range dependency. Variable 'x' used at line 146 is defined at line 145 and has a Short-Range dependency. Variable 'y' used at line 146 is defined at line 145 and has a Short-Range dependency. Library 'np' used at line 147 is imported at line 3 and has a Long-Range dependency. Variable 'output' used at line 147 is defined at line 137 and has a Short-Range dependency. Variable 'output' used at line 148 is defined at line 147 and has a Short-Range dependency.

{'Loop Body': 4}

{'Library Long-Range': 3, 'Variable Loop Short-Range': 4, 'Variable Short-Range': 10, 'Function Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):']

[' shift_center = np.array([j-center[0],i -center[1],1])']

[' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 143}]

Library 'np' used at line 143 is imported at line 3 and has a Long-Range dependency. Variable 'j' used at line 143 is part of a Loop defined at line 142 and has a Short-Range dependency. Variable 'center' used at line 143 is defined at line 138 and has a Short-Range dependency. Variable 'i' used at line 143 is part of a Loop defined at line 141 and has a Short-Range dependency.

{'Loop Body': 1}

{'Library Long-Range': 1, 'Variable Loop Short-Range': 2, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])']

[' coordinates = np.dot(T_invert,shift_center)']

[' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 144}]

Library 'np' used at line 144 is imported at line 3 and has a Long-Range dependency. Variable 'T_invert' used at line 144 is defined at line 139 and has a Short-Range dependency. Variable 'shift_center' used at line 144 is defined at line 143 and has a Short-Range dependency.

{'Loop Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)']

[' x,y = coordinates[0] + center[0], coordinates [1] + center[1]']

[' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 145}]

Variable 'coordinates' used at line 145 is defined at line 144 and has a Short-Range dependency. Variable 'center' used at line 145 is defined at line 138 and has a Short-Range dependency.

{'Loop Body': 1}

{'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]']

[' output[i][j] = bilinear_interpolation(I,x,y)']

[' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 146}]

Variable 'output' used at line 146 is defined at line 137 and has a Short-Range dependency. Variable 'i' used at line 146 is part of a Loop defined at line 141 and has a Short-Range dependency. Variable 'j' used at line 146 is part of a Loop defined at line 142 and has a Short-Range dependency. Function 'bilinear_interpolation' used at line 146 is defined at line 114 and has a Long-Range dependency. Variable 'I' used at line 146 is defined at line 135 and has a Medium-Range dependency. Variable 'x' used at line 146 is defined at line 145 and has a Short-Range dependency. Variable 'y' used at line 146 is defined at line 145 and has a Short-Range dependency.

{'Loop Body': 1}

{'Variable Short-Range': 3, 'Variable Loop Short-Range': 2, 'Function Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):']

[' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)']

[' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[{'reason_category': 'Loop Body', 'usage_line': 143}, {'reason_category': 'Loop Body', 'usage_line': 144}, {'reason_category': 'Loop Body', 'usage_line': 145}, {'reason_category': 'Loop Body', 'usage_line': 146}]

Library 'np' used at line 143 is imported at line 3 and has a Long-Range dependency. Variable 'j' used at line 143 is part of a Loop defined at line 142 and has a Short-Range dependency. Variable 'center' used at line 143 is defined at line 138 and has a Short-Range dependency. Variable 'i' used at line 143 is part of a Loop defined at line 141 and has a Short-Range dependency. Library 'np' used at line 144 is imported at line 3 and has a Long-Range dependency. Variable 'T_invert' used at line 144 is defined at line 139 and has a Short-Range dependency. Variable 'shift_center' used at line 144 is defined at line 143 and has a Short-Range dependency. Variable 'coordinates' used at line 145 is defined at line 144 and has a Short-Range dependency. Variable 'center' used at line 145 is defined at line 138 and has a Short-Range dependency. Variable 'output' used at line 146 is defined at line 137 and has a Short-Range dependency. Variable 'i' used at line 146 is part of a Loop defined at line 141 and has a Short-Range dependency. Variable 'j' used at line 146 is part of a Loop defined at line 142 and has a Short-Range dependency. Function 'bilinear_interpolation' used at line 146 is defined at line 114 and has a Long-Range dependency. Variable 'I' used at line 146 is defined at line 135 and has a Medium-Range dependency. Variable 'x' used at line 146 is defined at line 145 and has a Short-Range dependency. Variable 'y' used at line 146 is defined at line 145 and has a Short-Range dependency.

{'Loop Body': 4}

{'Library Long-Range': 2, 'Variable Loop Short-Range': 4, 'Variable Short-Range': 8, 'Function Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)']

['warped_arabella2 = image_warp(arabella_smol,t2)']

['t3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'image_warp' used at line 160 is defined at line 135 and has a Medium-Range dependency. Variable 'arabella_smol' used at line 160 is defined at line 154 and has a Short-Range dependency. Variable 't2' used at line 160 is defined at line 159 and has a Short-Range dependency.

{}

{'Function Medium-Range': 1, 'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)']

['warped_arabella4 = image_warp(arabella_smol,t4)']

['print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'image_warp' used at line 164 is defined at line 135 and has a Medium-Range dependency. Variable 'arabella_smol' used at line 164 is defined at line 154 and has a Short-Range dependency. Variable 't4' used at line 164 is defined at line 163 and has a Short-Range dependency.

{}

{'Function Medium-Range': 1, 'Variable Short-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise']

['r1 = rotation(30, False)', 'r2 = rotation(-30, False)']

['', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'rotation' used at line 171 is defined at line 86 and has a Long-Range dependency. Function 'rotation' used at line 172 is defined at line 86 and has a Long-Range dependency.

{}

{'Function Long-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)']

['r2 = rotation(-30, False)']

['', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'rotation' used at line 172 is defined at line 86 and has a Long-Range dependency.

{}

{'Function Long-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)']

['warped_arabella6 = image_warp(arabella_smol,r2)']

['print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'image_warp' used at line 175 is defined at line 135 and has a Long-Range dependency. Variable 'arabella_smol' used at line 175 is defined at line 154 and has a Medium-Range dependency. Variable 'r2' used at line 175 is defined at line 172 and has a Short-Range dependency.

{}

{'Function Long-Range': 1, 'Variable Medium-Range': 1, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)']

['warped_arabella7 = image_warp(arabella_smol,s1)']

['print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'image_warp' used at line 181 is defined at line 135 and has a Long-Range dependency. Variable 'arabella_smol' used at line 181 is defined at line 154 and has a Medium-Range dependency. Variable 's1' used at line 181 is defined at line 180 and has a Short-Range dependency.

{}

{'Function Long-Range': 1, 'Variable Medium-Range': 1, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)']

['warped_arabella8 = image_warp(arabella_smol,a1)']

['print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Function 'image_warp' used at line 186 is defined at line 135 and has a Long-Range dependency. Variable 'arabella_smol' used at line 186 is defined at line 154 and has a Long-Range dependency. Variable 'a1' used at line 186 is defined at line 185 and has a Short-Range dependency.

{}

{'Function Long-Range': 1, 'Variable Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '']

['synthI12 = I1+I2']

['diffI = synthI12 - I12', 'diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'I1' used at line 207 is defined at line 195 and has a Medium-Range dependency. Variable 'I2' used at line 207 is defined at line 196 and has a Medium-Range dependency.

{}

{'Variable Medium-Range': 2}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2']

['diffI = synthI12 - I12']

['diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))', '', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'synthI12' used at line 208 is defined at line 207 and has a Short-Range dependency. Variable 'I12' used at line 208 is defined at line 197 and has a Medium-Range dependency.

{}

{'Variable Short-Range': 1, 'Variable Medium-Range': 1}

infilling_python

Image_Transformation

['from PIL import Image', 'import cv2', 'import numpy as np', 'import matplotlib.pyplot as plt', '', '#Task 1:', '#resize two imgs', '#resize imgA with cv2.resize', '#resize imgB via center cropping', '#ensure both resized images are resized to same end_size', '#concatenate the resized imgs via every other row imgA and every other row imgB', '', "A = Image.open('./imgA.jpg')", "B = Image.open('./imgB.jpg')", 'A_array = np.array(A)', 'B_array = np.array(B)', "print('Array shapes:',A_array.shape, B_array.shape)", 'print(A_array)', 'print(B_array)', '', 'end_size = 256', '', 'if end_size >= A_array.shape[0] or end_size>= B_array.shape[0]:', " print('choose end size less than: ', np.min(A_array.shape,B_array.shape))", '', 'A_resized = cv2.resize(A_array, (end_size, end_size))', '', 'def center_crop(img_array, end_size):', ' x_start = int((img_array.shape[0]-end_size)/2)', ' x_end = x_start + end_size', '', ' y_start = int((img_array.shape[1]-end_size)/2)', ' y_end = y_start + end_size', '', ' img_resized = img_array[x_start:x_end, y_start:y_end, :]', ' return img_resized', 'B_resized = center_crop(B_array, end_size)', 'print(B_resized.shape)', 'C = np.concatenate((A_resized[0:256,0:128,:],B_resized[0:256,128:256,:]),axis = 1)', '', 'D = B_resized[0:1,:,:]', 'for row in range(1,A_resized.shape[0]):', ' if row % 2 == 0:', ' D = np.concatenate((D,B_resized[row:row+1,:,:]), axis=0)', ' else:', ' D = np.concatenate((D,A_resized[row:row+1,:,:]), axis =0)', 'print(D)', '', '#Task 2:', '#upload picture of multiple peppers each different colors', '#create a mask without the yellow peppers by using range provided', '#do this in rgb and hsv -> note different ranges depending on rgb or hsv', "pepper_img = Image.open('./pepper.png')", 'pepper = np.array(pepper_img)', '', 'lower_yellow = np.array([150, 175, 0], dtype=np.uint8)', 'upper_yellow = np.array([255, 255, 150], dtype=np.uint8)', 'mask = np.all((pepper >= lower_yellow) & (pepper <= upper_yellow), axis=-1)', '', 'result = np.where(mask, 1, 0)', '', 'print(pepper)', 'print(result)', '', "img = cv2.imread('./pepper.png')", 'hsv_img = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)', '', 'lower_yellow = np.array([19, 0, 0], dtype=np.uint8)', 'upper_yellow = np.array([24, 255, 255], dtype=np.uint8)', 'mask = np.all((hsv_img >= lower_yellow) & (hsv_img <= upper_yellow), axis=-1)', '', 'result_hsv = np.where(mask, 1, 0)', '', 'print(hsv_img)', 'print(result_hsv)', '', '#Task 3:', '#write transormation functions to translsate, rotate, and perform similarity and affine transformation', '#write bilinear interpolation function from scratch', '#apply series of transformations to an image', '', 'def translation(dx,dy):', ' translation_matrix = np.array([[1,0,dx],[0,1,dy],[0,0,1]])', ' return translation_matrix', '', 'def rotation(angle,radians = True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', ' rotation_matrix = np.array([[costheta, sintheta,0],[-1*sintheta,costheta,0],[0,0,1]])', ' return rotation_matrix', '', 'def similarity_matrix(angle, dx, dy, scale_factor,radians=True):', ' if radians == False:', ' angle = np.radians(angle)', ' costheta = np.cos(angle)', ' sintheta = np.sin(angle)', '', ' similarity_matrix = np.array([[scale_factor*costheta,scale_factor*sintheta,dx],', ' [-1*scale_factor*sintheta, scale_factor*costheta, dy],', ' [0,0,1]])', ' return similarity_matrix', '', 'def affine(angle, x, y, scale, ax, ay):', ' scaling = np.array([[scale, 0,0], [0, scale, 0], [0,0,1]])', ' shear = np.array([[1, ax, 0], [ay, 1,0], [0, 0,1]])', ' result = np.array([[0,0,0], [0,0,0], [0,0,0]])', ' result = np.dot(translation(x, y), rotation(angle))', ' result = np.dot(result, scaling)', ' result = np.dot(result, shear)', ' return result', '', 'def bilinear_interpolation(image,x,y):', ' x1 = int(x)', ' x2 = x1 + 1', ' y1 = int(y)', ' y2 = y1 + 1', '', ' if x1 < 0 or y1 < 0 or x2 >= image.shape[1] or y2 >= image.shape[0]:', ' return 0', ' else:', ' f11 = image[y1][x1]', ' f12 = image[y1][x2]', ' f21 = image[y2][x1]', ' f22 = image[y2][x2]', '', ' w1 = (x2-x)*(y2-y)', ' w2 = (x-x1)*(y2-y)', ' w3 = (x2-x)*(y-y1)', ' w4 = (x-x1)*(y-y1)', '', ' return (w1*f11) + (w2*f12) + (w3*f21) + (w4*f22)', '', 'def image_warp(I,T):', ' rows,cols = I.shape[:2]', ' output = np.zeros((rows,cols,3))', ' center = (cols/2, rows/2)', ' T_invert = np.linalg.inv(T)', '', ' for i in range(rows):', ' for j in range(cols):', ' shift_center = np.array([j-center[0],i -center[1],1])', ' coordinates = np.dot(T_invert,shift_center)', ' x,y = coordinates[0] + center[0], coordinates [1] + center[1]', ' output[i][j] = bilinear_interpolation(I,x,y)', ' output = np.array(output, np.uint8)', ' return output', '', "path= './arabella.jpg'", 'arabella = cv2.imread(path)', 'arabella_smol = cv2.resize(arabella, dsize=(256, 192), interpolation=cv2.INTER_AREA)', 'arabella_smol = np.array(arabella_smol)', 'arabella_smol = arabella_smol[:, :, [2, 1, 0]]', '', '#translate images keep params as shown', 't1 = translation(21,25)', 'warped_arabella1 = image_warp(arabella_smol,t1)', 't2 = translation(-21,25)', 'warped_arabella2 = image_warp(arabella_smol,t2)', 't3 = translation(21,-25)', 'warped_arabella3 = image_warp(arabella_smol,t3)', 't4 = translation(-21,25)', 'warped_arabella4 = image_warp(arabella_smol,t4)', 'print(warped_arabella1)', 'print(warped_arabella2)', 'print(warped_arabella3)', 'print(warped_arabella4)', '', '# rotate image 30 degrees clockwise and 30 degrees counterclockwise', 'r1 = rotation(30, False)', 'r2 = rotation(-30, False)', '', 'warped_arabella5 = image_warp(arabella_smol,r1)', 'warped_arabella6 = image_warp(arabella_smol,r2)', 'print(warped_arabella5)', 'print(warped_arabella6)', '', '#apply similarity transformation to image, keep params as shown below', 's1 = similarity_matrix(60, 0, 0, 0.5,radians=False)', 'warped_arabella7 = image_warp(arabella_smol,s1)', 'print(warped_arabella7)', '', '#apply affine transformation to image, keep params as shown below', 'a1 = affine(90, 2, 3, .5, 5, 2)', 'warped_arabella8 = image_warp(arabella_smol,a1)', 'print(warped_arabella8)', '', '#Task 4:', '#artificially replicate overhead and desklight scene via addition of overhead lit only and desklight lit only scenes', '#make sure to scale properly ', "path1= './desklight.jpg'", "path2= './overheadlight.jpg'", "path3 = './bothlight.jpg'", 'I1 = np.array(cv2.imread(path1))[:,:,[2,1,0]]', 'I2 = np.array(cv2.imread(path2))[:,:,[2,1,0]]', 'I12 = np.array(cv2.imread(path3))[:,:,[2,1,0]]', '', 'type(I12[0,0,0])', 'I1_float = I1/255.0', 'I2_float = I2/255.0', 'I12_float = I1_float + I2_float', 'type(I12_float[0,0,0]),np.min(I12_float),np.max(I12_float)', 'I12_uint8 = (I12_float * 255.0).astype(np.uint8)', 'type(I12_uint8[0,0,0]),np.min(I12_uint8),np.max(I12_uint8)', '', 'synthI12 = I1+I2', 'diffI = synthI12 - I12']

['diffI_scaled = (diffI - np.min(diffI))/(np.max(diffI)-np.min(diffI))']

['', 'print(I12)', 'print(synthI12)', 'print(diffI_scaled)']

[]

Variable 'diffI' used at line 209 is defined at line 208 and has a Short-Range dependency. Library 'np' used at line 209 is imported at line 3 and has a Long-Range dependency.

{}

{'Variable Short-Range': 1, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module']

['script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)']

['', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'os' used at line 43 is imported at line 8 and has a Long-Range dependency. Library 'os' used at line 44 is imported at line 8 and has a Long-Range dependency. Variable 'script_path' used at line 44 is defined at line 43 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):']

[' os.makedirs(log_dir, exist_ok=True)']

['', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 48}]

Library 'os' used at line 48 is imported at line 8 and has a Long-Range dependency. Variable 'log_dir' used at line 48 is defined at line 46 and has a Short-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals']

[' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)']

[' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Lambda_Expressions', 'usage_line': 83}]

Library 'tf' used at line 83 is imported at line 20 and has a Long-Range dependency. Variable 'x' used at line 83 is part of a Lambda_Expressions defined at line 83 and has a Short-Range dependency. Variable 'num_objs' used at line 83 is defined at line 74 and has a Short-Range dependency. Variable 'states' used at line 83 is defined at line 81 and has a Short-Range dependency.

{'Lambda_Expressions': 1}

{'Library Long-Range': 1, 'Variable Lambda_Expressions Short-Range': 1, 'Variable Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq']

[' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq']

[' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Condition', 'usage_line': 104}, {'reason_category': 'Elif Body', 'usage_line': 105}, {'reason_category': 'Elif Body', 'usage_line': 106}, {'reason_category': 'Elif Body', 'usage_line': 107}, {'reason_category': 'Elif Body', 'usage_line': 108}, {'reason_category': 'Elif Body', 'usage_line': 109}, {'reason_category': 'Elif Body', 'usage_line': 110}, {'reason_category': 'Elif Condition', 'usage_line': 111}, {'reason_category': 'Elif Body', 'usage_line': 112}, {'reason_category': 'Elif Body', 'usage_line': 113}, {'reason_category': 'Elif Body', 'usage_line': 114}, {'reason_category': 'Elif Body', 'usage_line': 115}, {'reason_category': 'Elif Body', 'usage_line': 116}, {'reason_category': 'Elif Body', 'usage_line': 117}]

Variable 'args' used at line 104 is defined at line 89 and has a Medium-Range dependency. Library 'tf' used at line 106 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 106 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 107 is defined at line 106 and has a Short-Range dependency. Variable 'skill_seq' used at line 109 is defined at line 107 and has a Short-Range dependency. Library 'tf' used at line 109 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 110 is defined at line 109 and has a Short-Range dependency. Variable 'args' used at line 111 is defined at line 89 and has a Medium-Range dependency. Library 'tf' used at line 113 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 113 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 114 is defined at line 113 and has a Short-Range dependency. Variable 'skill_seq' used at line 116 is defined at line 114 and has a Short-Range dependency. Library 'tf' used at line 116 is imported at line 20 and has a Long-Range dependency. Variable 'args' used at line 116 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 117 is defined at line 116 and has a Short-Range dependency.

{'Elif Condition': 2, 'Elif Body': 12}

{'Variable Medium-Range': 5, 'Library Long-Range': 4, 'Variable Short-Range': 6}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":']

[' return skill_seq']

[' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 103}]

Variable 'skill_seq' used at line 103 is defined at line 89 and has a Medium-Range dependency.

{'If Body': 1}

{'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq']

[' elif args.wrap_level == "1":']

[' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Condition', 'usage_line': 104}]

Variable 'args' used at line 104 is defined at line 89 and has a Medium-Range dependency.

{'Elif Condition': 1}

{'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3']

[' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3']

[' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Body', 'usage_line': 106}, {'reason_category': 'Elif Body', 'usage_line': 107}]

Library 'tf' used at line 106 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 106 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 107 is defined at line 106 and has a Short-Range dependency.

{'Elif Body': 2}

{'Library Long-Range': 1, 'Variable Medium-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot']

[' skill_seq = tf.one_hot(skill_seq, depth=3)']

[' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Body', 'usage_line': 109}]

Variable 'skill_seq' used at line 109 is defined at line 107 and has a Short-Range dependency. Library 'tf' used at line 109 is imported at line 20 and has a Long-Range dependency.

{'Elif Body': 1}

{'Variable Short-Range': 1, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq']

[' elif args.wrap_level == "2":']

[' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Condition', 'usage_line': 111}]

Variable 'args' used at line 111 is defined at line 89 and has a Medium-Range dependency.

{'Elif Condition': 1}

{'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs']

[' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3']

[' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Body', 'usage_line': 113}, {'reason_category': 'Elif Body', 'usage_line': 114}]

Library 'tf' used at line 113 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 113 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 114 is defined at line 113 and has a Short-Range dependency.

{'Elif Body': 2}

{'Library Long-Range': 1, 'Variable Medium-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot']

[' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)']

[' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Body', 'usage_line': 116}]

Variable 'skill_seq' used at line 116 is defined at line 114 and has a Short-Range dependency. Library 'tf' used at line 116 is imported at line 20 and has a Long-Range dependency. Variable 'args' used at line 116 is defined at line 89 and has a Medium-Range dependency.

{'Elif Body': 1}

{'Variable Short-Range': 1, 'Library Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":']

[' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq']

[' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 103}, {'reason_category': 'Elif Condition', 'usage_line': 104}, {'reason_category': 'Elif Body', 'usage_line': 105}, {'reason_category': 'Elif Body', 'usage_line': 106}, {'reason_category': 'Elif Body', 'usage_line': 107}, {'reason_category': 'Elif Body', 'usage_line': 108}, {'reason_category': 'Elif Body', 'usage_line': 109}, {'reason_category': 'Elif Body', 'usage_line': 110}, {'reason_category': 'Elif Condition', 'usage_line': 111}, {'reason_category': 'Elif Body', 'usage_line': 112}, {'reason_category': 'Elif Body', 'usage_line': 113}, {'reason_category': 'Elif Body', 'usage_line': 114}, {'reason_category': 'Elif Body', 'usage_line': 115}, {'reason_category': 'Elif Body', 'usage_line': 116}, {'reason_category': 'Elif Body', 'usage_line': 117}]

Variable 'skill_seq' used at line 103 is defined at line 89 and has a Medium-Range dependency. Variable 'args' used at line 104 is defined at line 89 and has a Medium-Range dependency. Library 'tf' used at line 106 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 106 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 107 is defined at line 106 and has a Short-Range dependency. Variable 'skill_seq' used at line 109 is defined at line 107 and has a Short-Range dependency. Library 'tf' used at line 109 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 110 is defined at line 109 and has a Short-Range dependency. Variable 'args' used at line 111 is defined at line 89 and has a Medium-Range dependency. Library 'tf' used at line 113 is imported at line 20 and has a Long-Range dependency. Variable 'skill_seq' used at line 113 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 114 is defined at line 113 and has a Short-Range dependency. Variable 'skill_seq' used at line 116 is defined at line 114 and has a Short-Range dependency. Library 'tf' used at line 116 is imported at line 20 and has a Long-Range dependency. Variable 'args' used at line 116 is defined at line 89 and has a Medium-Range dependency. Variable 'skill_seq' used at line 117 is defined at line 116 and has a Short-Range dependency.

{'If Body': 1, 'Elif Condition': 2, 'Elif Body': 12}

{'Variable Medium-Range': 6, 'Library Long-Range': 4, 'Variable Short-Range': 6}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """']

[' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef']

['', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 133}, {'reason_category': 'Loop Body', 'usage_line': 134}, {'reason_category': 'If Condition', 'usage_line': 134}, {'reason_category': 'Loop Body', 'usage_line': 135}, {'reason_category': 'If Body', 'usage_line': 135}, {'reason_category': 'Loop Body', 'usage_line': 136}, {'reason_category': 'If Body', 'usage_line': 136}]

Variable 'model' used at line 133 is defined at line 122 and has a Medium-Range dependency. Variable 'layer' used at line 134 is part of a Loop defined at line 133 and has a Short-Range dependency. Library 'tf' used at line 134 is imported at line 20 and has a Long-Range dependency. Library 'tf' used at line 135 is imported at line 20 and has a Long-Range dependency. Variable 'layer' used at line 135 is part of a Loop defined at line 133 and has a Short-Range dependency. Variable 'reg' used at line 136 is defined at line 132 and has a Short-Range dependency. Library 'tf' used at line 136 is imported at line 20 and has a Long-Range dependency. Variable 'prod' used at line 136 is defined at line 135 and has a Short-Range dependency. Variable 'reg' used at line 138 is defined at line 132 and has a Short-Range dependency. Variable 'reg_coef' used at line 138 is defined at line 122 and has a Medium-Range dependency. Variable 'reg' used at line 139 is defined at line 132 and has a Short-Range dependency. Variable 'reg_coef' used at line 139 is defined at line 122 and has a Medium-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 3, 'If Condition': 1, 'If Body': 2}

{'Variable Medium-Range': 3, 'Variable Loop Short-Range': 2, 'Library Long-Range': 3, 'Variable Short-Range': 4}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:']

[' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef']

['', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Loop Body', 'usage_line': 134}, {'reason_category': 'If Condition', 'usage_line': 134}, {'reason_category': 'Loop Body', 'usage_line': 135}, {'reason_category': 'If Body', 'usage_line': 135}, {'reason_category': 'Loop Body', 'usage_line': 136}, {'reason_category': 'If Body', 'usage_line': 136}]

Variable 'layer' used at line 134 is part of a Loop defined at line 133 and has a Short-Range dependency. Library 'tf' used at line 134 is imported at line 20 and has a Long-Range dependency. Library 'tf' used at line 135 is imported at line 20 and has a Long-Range dependency. Variable 'layer' used at line 135 is part of a Loop defined at line 133 and has a Short-Range dependency. Variable 'reg' used at line 136 is defined at line 132 and has a Short-Range dependency. Library 'tf' used at line 136 is imported at line 20 and has a Long-Range dependency. Variable 'prod' used at line 136 is defined at line 135 and has a Short-Range dependency. Variable 'reg' used at line 138 is defined at line 132 and has a Short-Range dependency. Variable 'reg_coef' used at line 138 is defined at line 122 and has a Medium-Range dependency. Variable 'reg' used at line 139 is defined at line 132 and has a Short-Range dependency. Variable 'reg_coef' used at line 139 is defined at line 122 and has a Medium-Range dependency.

{'Loop Body': 3, 'If Condition': 1, 'If Body': 2}

{'Variable Loop Short-Range': 2, 'Library Long-Range': 3, 'Variable Short-Range': 4, 'Variable Medium-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:']

[' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))']

[' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Loop Body', 'usage_line': 134}, {'reason_category': 'If Condition', 'usage_line': 134}, {'reason_category': 'Loop Body', 'usage_line': 135}, {'reason_category': 'If Body', 'usage_line': 135}, {'reason_category': 'Loop Body', 'usage_line': 136}, {'reason_category': 'If Body', 'usage_line': 136}]

Variable 'layer' used at line 134 is part of a Loop defined at line 133 and has a Short-Range dependency. Library 'tf' used at line 134 is imported at line 20 and has a Long-Range dependency. Library 'tf' used at line 135 is imported at line 20 and has a Long-Range dependency. Variable 'layer' used at line 135 is part of a Loop defined at line 133 and has a Short-Range dependency. Variable 'reg' used at line 136 is defined at line 132 and has a Short-Range dependency. Library 'tf' used at line 136 is imported at line 20 and has a Long-Range dependency. Variable 'prod' used at line 136 is defined at line 135 and has a Short-Range dependency.

{'Loop Body': 3, 'If Condition': 1, 'If Body': 2}

{'Variable Loop Short-Range': 2, 'Library Long-Range': 3, 'Variable Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))']

[' return reg * reg_coef']

['', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'reg' used at line 139 is defined at line 132 and has a Short-Range dependency. Variable 'reg_coef' used at line 139 is defined at line 122 and has a Medium-Range dependency.

{}

{'Variable Short-Range': 1, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated']

[' terminate_idxes = tf.math.argmax(successes, axis=-1)']

[' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 154 is imported at line 20 and has a Long-Range dependency. Variable 'successes' used at line 154 is defined at line 151 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index']

[' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),']

[' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 156 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 156 is defined at line 154 and has a Short-Range dependency. Variable 'terminate_idxes' used at line 157 is defined at line 154 and has a Short-Range dependency. Library 'tf' used at line 157 is imported at line 20 and has a Long-Range dependency. Variable 'T' used at line 157 is defined at line 150 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success"]

[' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)']

[' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'terminate_idxes' used at line 161 is defined at line 157 and has a Short-Range dependency. Library 'tf' used at line 161 is imported at line 20 and has a Long-Range dependency. Library 'tfp' used at line 162 is imported at line 21 and has a Long-Range dependency. Variable 'p' used at line 162 is defined at line 161 and has a Short-Range dependency. Variable 'batch_size' used at line 162 is defined at line 149 and has a Medium-Range dependency. Library 'tf' used at line 163 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 163 is defined at line 157 and has a Short-Range dependency. Variable 'episode_idxs' used at line 163 is defined at line 162 and has a Short-Range dependency.

{}

{'Variable Short-Range': 4, 'Library Long-Range': 3, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)']

[' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)']

[' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tfp' used at line 162 is imported at line 21 and has a Long-Range dependency. Variable 'p' used at line 162 is defined at line 161 and has a Short-Range dependency. Variable 'batch_size' used at line 162 is defined at line 149 and has a Medium-Range dependency. Library 'tf' used at line 163 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 163 is defined at line 157 and has a Short-Range dependency. Variable 'episode_idxs' used at line 163 is defined at line 162 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 3, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes']

[' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)']

[' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 165 is imported at line 20 and has a Long-Range dependency. Variable 'episode_idxs' used at line 165 is defined at line 163 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step']

[' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)']

[' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 171 is imported at line 20 and has a Long-Range dependency. Variable 'batch_size' used at line 171 is defined at line 149 and has a Medium-Range dependency. Variable 't_samples_frac' used at line 172 is defined at line 171 and has a Short-Range dependency. Library 'tf' used at line 172 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 172 is defined at line 165 and has a Short-Range dependency. Library 'tf' used at line 173 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 173 is defined at line 165 and has a Short-Range dependency.

{}

{'Library Long-Range': 3, 'Variable Medium-Range': 1, 'Variable Short-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)']

[' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)']

[' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 176 is imported at line 20 and has a Long-Range dependency. Variable 't_samples_frac' used at line 176 is defined at line 171 and has a Short-Range dependency. Variable 'rdm_past_offset_frac' used at line 177 is defined at line 176 and has a Short-Range dependency. Library 'tf' used at line 177 is imported at line 20 and has a Long-Range dependency. Variable 't_samples' used at line 177 is defined at line 173 and has a Short-Range dependency. Library 'tf' used at line 178 is imported at line 20 and has a Long-Range dependency. Variable 't_samples' used at line 178 is defined at line 173 and has a Short-Range dependency.

{}

{'Library Long-Range': 3, 'Variable Short-Range': 4}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)']

[' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)']

[' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'rdm_past_offset_frac' used at line 177 is defined at line 176 and has a Short-Range dependency. Library 'tf' used at line 177 is imported at line 20 and has a Long-Range dependency. Variable 't_samples' used at line 177 is defined at line 173 and has a Short-Range dependency. Library 'tf' used at line 178 is imported at line 20 and has a Long-Range dependency. Variable 't_samples' used at line 178 is defined at line 173 and has a Short-Range dependency.

{}

{'Variable Short-Range': 3, 'Library Long-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step']

[' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)']

[' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 182 is imported at line 20 and has a Long-Range dependency. Variable 'batch_size' used at line 182 is defined at line 149 and has a Long-Range dependency. Variable 'rdm_future_offset_frac' used at line 183 is defined at line 182 and has a Short-Range dependency. Library 'tf' used at line 183 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 183 is defined at line 165 and has a Medium-Range dependency. Variable 't_samples' used at line 183 is defined at line 173 and has a Short-Range dependency. Library 'tf' used at line 184 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 184 is defined at line 165 and has a Medium-Range dependency. Variable 'future_offset' used at line 184 is defined at line 183 and has a Short-Range dependency.

{}

{'Library Long-Range': 3, 'Variable Long-Range': 1, 'Variable Short-Range': 3, 'Variable Medium-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step']

[' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset']

[' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 182 is imported at line 20 and has a Long-Range dependency. Variable 'batch_size' used at line 182 is defined at line 149 and has a Long-Range dependency. Variable 'rdm_future_offset_frac' used at line 183 is defined at line 182 and has a Short-Range dependency. Library 'tf' used at line 183 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 183 is defined at line 165 and has a Medium-Range dependency. Variable 't_samples' used at line 183 is defined at line 173 and has a Short-Range dependency. Library 'tf' used at line 184 is imported at line 20 and has a Long-Range dependency. Variable 'terminate_idxes' used at line 184 is defined at line 165 and has a Medium-Range dependency. Variable 'future_offset' used at line 184 is defined at line 183 and has a Short-Range dependency. Variable 't_samples' used at line 185 is defined at line 173 and has a Medium-Range dependency. Variable 'future_offset' used at line 185 is defined at line 184 and has a Short-Range dependency.

{}

{'Library Long-Range': 3, 'Variable Long-Range': 1, 'Variable Short-Range': 4, 'Variable Medium-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------']

[' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)']

[' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 192}, {'reason_category': 'Loop Body', 'usage_line': 193}]

Library 'tf' used at line 190 is imported at line 20 and has a Long-Range dependency. Variable 'episode_idxs' used at line 190 is defined at line 163 and has a Medium-Range dependency. Variable 't_samples' used at line 190 is defined at line 173 and has a Medium-Range dependency. Variable 'episodic_data' used at line 192 is defined at line 143 and has a Long-Range dependency. Variable 'transitions' used at line 193 is defined at line 191 and has a Short-Range dependency. Variable 'key' used at line 193 is part of a Loop defined at line 192 and has a Short-Range dependency. Library 'tf' used at line 193 is imported at line 20 and has a Long-Range dependency. Variable 'episodic_data' used at line 193 is defined at line 143 and has a Long-Range dependency. Variable 'curr_indices' used at line 193 is defined at line 190 and has a Short-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 1}

{'Library Long-Range': 2, 'Variable Medium-Range': 2, 'Variable Long-Range': 2, 'Variable Short-Range': 2, 'Variable Loop Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}']

[' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)']

[' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 192}, {'reason_category': 'Loop Body', 'usage_line': 193}]

Variable 'episodic_data' used at line 192 is defined at line 143 and has a Long-Range dependency. Variable 'transitions' used at line 193 is defined at line 191 and has a Short-Range dependency. Variable 'key' used at line 193 is part of a Loop defined at line 192 and has a Short-Range dependency. Library 'tf' used at line 193 is imported at line 20 and has a Long-Range dependency. Variable 'episodic_data' used at line 193 is defined at line 143 and has a Long-Range dependency. Variable 'curr_indices' used at line 193 is defined at line 190 and has a Short-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 1}

{'Variable Long-Range': 2, 'Variable Short-Range': 2, 'Variable Loop Short-Range': 1, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal("]

[" states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)']

[' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 196 is imported at line 20 and has a Long-Range dependency. Variable 'episodic_data' used at line 196 is defined at line 143 and has a Long-Range dependency. Variable 'curr_indices' used at line 196 is defined at line 190 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------']

[' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)']

[" transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 200 is imported at line 20 and has a Long-Range dependency. Variable 'episode_idxs' used at line 200 is defined at line 163 and has a Long-Range dependency. Variable 't_samples_future' used at line 200 is defined at line 185 and has a Medium-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)']

[" transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER']

[' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'transitions' used at line 201 is defined at line 195 and has a Short-Range dependency. Variable 'state_to_goal' used at line 201 is defined at line 142 and has a Long-Range dependency. Library 'tf' used at line 201 is imported at line 20 and has a Long-Range dependency. Variable 'episodic_data' used at line 201 is defined at line 143 and has a Long-Range dependency. Variable 'future_indices' used at line 201 is defined at line 200 and has a Short-Range dependency.

{}

{'Variable Short-Range': 2, 'Variable Long-Range': 2, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------']

[' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)']

[" transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 205 is imported at line 20 and has a Long-Range dependency. Variable 'episode_idxs' used at line 205 is defined at line 163 and has a Long-Range dependency. Variable 't_samples_init' used at line 205 is defined at line 178 and has a Medium-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Long-Range': 1, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)']

[" transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)"]

[' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'transitions' used at line 206 is defined at line 201 and has a Short-Range dependency. Library 'tf' used at line 206 is imported at line 20 and has a Long-Range dependency. Variable 'episodic_data' used at line 206 is defined at line 143 and has a Long-Range dependency. Variable 'init_indices' used at line 206 is defined at line 205 and has a Short-Range dependency.

{}

{'Variable Short-Range': 2, 'Library Long-Range': 1, 'Variable Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':"]

[' return sample_random_transitions']

[' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 211}]

Function 'sample_random_transitions' used at line 211 is defined at line 143 and has a Long-Range dependency.

{'If Body': 1}

{'Function Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))']

[' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]']

[' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 243}, {'reason_category': 'Loop Body', 'usage_line': 244}]

Variable 'self' used at line 243 is defined at line 239 and has a Short-Range dependency. Variable 'buffered_data' used at line 244 is defined at line 241 and has a Short-Range dependency. Variable 'key' used at line 244 is part of a Loop defined at line 243 and has a Short-Range dependency. Variable '_data' used at line 244 is defined at line 242 and has a Short-Range dependency. Variable 'index' used at line 244 is part of a Loop defined at line 243 and has a Short-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 1}

{'Variable Short-Range': 3, 'Variable Loop Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ']

[' transitions = self.transition_fn(buffered_data, batch_size)']

[' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 246 is defined at line 239 and has a Short-Range dependency. Variable 'buffered_data' used at line 246 is defined at line 241 and has a Short-Range dependency. Variable 'batch_size' used at line 246 is defined at line 239 and has a Short-Range dependency.

{}

{'Variable Short-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:']

[' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)']

[' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 255}]

Library 'tf' used at line 255 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 255 is defined at line 251 and has a Short-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:']

[' num_episodes = self.current_size']

[' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 257}, {'reason_category': 'Else Reasoning', 'usage_line': 257}]

Variable 'self' used at line 257 is defined at line 251 and has a Short-Range dependency.

{'If Body': 1, 'Else Reasoning': 1}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size']

[' ep_range = tf.range(num_episodes)']

[' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 258}]

Library 'tf' used at line 258 is imported at line 20 and has a Long-Range dependency. Variable 'num_episodes' used at line 258 is defined at line 257 and has a Short-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:']

[' ep_range = tf.range(ep_start, ep_end)']

[' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Else Reasoning', 'usage_line': 260}]

Library 'tf' used at line 260 is imported at line 20 and has a Long-Range dependency. Variable 'ep_start' used at line 260 is defined at line 251 and has a Short-Range dependency. Variable 'ep_end' used at line 260 is defined at line 251 and has a Short-Range dependency.

{'Else Reasoning': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)']

[' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]']

[' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 264}, {'reason_category': 'Loop Body', 'usage_line': 265}]

Variable 'self' used at line 264 is defined at line 251 and has a Medium-Range dependency. Variable 'buffered_data' used at line 265 is defined at line 262 and has a Short-Range dependency. Variable 'key' used at line 265 is part of a Loop defined at line 264 and has a Short-Range dependency. Variable '_data' used at line 265 is defined at line 263 and has a Short-Range dependency. Variable 'index' used at line 265 is part of a Loop defined at line 264 and has a Short-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 1}

{'Variable Medium-Range': 1, 'Variable Short-Range': 2, 'Variable Loop Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)']

[' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)']

[' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 278 is defined at line 270 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}']

[' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)']

[' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Loop Body', 'usage_line': 283}, {'reason_category': 'Define Stop Criteria', 'usage_line': 283}, {'reason_category': 'Loop Body', 'usage_line': 284}, {'reason_category': 'Loop Body', 'usage_line': 285}]

Variable 'self' used at line 283 is defined at line 280 and has a Short-Range dependency. Variable 'episode_batch' used at line 284 is defined at line 282 and has a Short-Range dependency. Variable 'key' used at line 284 is part of a Loop defined at line 283 and has a Short-Range dependency. Library 'tf' used at line 284 is imported at line 20 and has a Long-Range dependency. Variable 'episodes_batch' used at line 284 is defined at line 280 and has a Short-Range dependency. Variable 'ep_idx' used at line 284 is part of a Loop defined at line 281 and has a Short-Range dependency. Variable 'self' used at line 285 is defined at line 280 and has a Short-Range dependency. Variable 'episode_batch' used at line 285 is defined at line 282 and has a Short-Range dependency.

{'Loop Body': 3, 'Define Stop Criteria': 1}

{'Variable Short-Range': 5, 'Variable Loop Short-Range': 2, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:']

[' idxs = tf.range(self.current_size, self.current_size + num_to_ins)']

[' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 293}]

Library 'tf' used at line 293 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 293 is defined at line 287 and has a Short-Range dependency. Variable 'num_to_ins' used at line 293 is defined at line 287 and has a Short-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:']

[' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)']

[' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Elif Body', 'usage_line': 295}, {'reason_category': 'Elif Body', 'usage_line': 296}, {'reason_category': 'Elif Body', 'usage_line': 297}, {'reason_category': 'Elif Body', 'usage_line': 298}]

Variable 'num_to_ins' used at line 295 is defined at line 287 and has a Short-Range dependency. Variable 'self' used at line 295 is defined at line 287 and has a Short-Range dependency. Library 'tf' used at line 296 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 296 is defined at line 287 and has a Short-Range dependency. Library 'tf' used at line 297 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 297 is defined at line 287 and has a Short-Range dependency. Variable 'overflow' used at line 297 is defined at line 295 and has a Short-Range dependency. Library 'tf' used at line 298 is imported at line 20 and has a Long-Range dependency. Variable 'idx_a' used at line 298 is defined at line 296 and has a Short-Range dependency. Variable 'idx_b' used at line 298 is defined at line 297 and has a Short-Range dependency.

{'Elif Body': 4}

{'Variable Short-Range': 7, 'Library Long-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:']

[' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)']

[' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Else Reasoning', 'usage_line': 300}]

Library 'tf' used at line 300 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 300 is defined at line 287 and has a Medium-Range dependency. Variable 'num_to_ins' used at line 300 is defined at line 287 and has a Medium-Range dependency.

{'Else Reasoning': 1}

{'Library Long-Range': 1, 'Variable Medium-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size']

[' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))']

[' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 303 is defined at line 287 and has a Medium-Range dependency. Library 'tf' used at line 303 is imported at line 20 and has a Long-Range dependency. Variable 'num_to_ins' used at line 303 is defined at line 287 and has a Medium-Range dependency.

{}

{'Variable Medium-Range': 2, 'Library Long-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):']

[' return self.current_size']

[' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 308 is defined at line 307 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):']

[' return self.current_size * self.T']

[' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 311 is defined at line 310 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):']

[' self.current_size.assign(0)']

[' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 314 is defined at line 313 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):']

[' return self.current_size == self.buffer_size']

[' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 318 is defined at line 317 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):']

[' return self.current_size']

[' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 321 is defined at line 320 and has a Short-Range dependency.

{}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}']

[' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]']

[' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'Define Stop Criteria', 'usage_line': 326}, {'reason_category': 'Loop Body', 'usage_line': 327}]

Variable 'self' used at line 325 is defined at line 323 and has a Short-Range dependency. Library 'tf' used at line 325 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 326 is defined at line 323 and has a Short-Range dependency. Variable 'buffered_data' used at line 327 is defined at line 324 and has a Short-Range dependency. Variable 'key' used at line 327 is part of a Loop defined at line 326 and has a Short-Range dependency. Variable '_data' used at line 327 is defined at line 325 and has a Short-Range dependency. Variable 'index' used at line 327 is part of a Loop defined at line 326 and has a Short-Range dependency.

{'Define Stop Criteria': 1, 'Loop Body': 1}

{'Variable Short-Range': 4, 'Library Long-Range': 1, 'Variable Loop Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:']

[' self.clear_buffer()']

[' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 338}]

Variable 'self' used at line 338 is defined at line 332 and has a Short-Range dependency.

{'If Body': 1}

{'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load']

[" idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()"]

[' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 343}]

Library 'np' used at line 343 is imported at line 19 and has a Long-Range dependency. Variable 'buffered_data' used at line 343 is defined at line 332 and has a Medium-Range dependency. Variable 'num_demos_to_load' used at line 343 is defined at line 332 and has a Medium-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Medium-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data']

[' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])']

[' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'List_Comprehension', 'usage_line': 349}]

Variable 'key' used at line 349 is part of a List_Comprehension defined at line 349 and has a Short-Range dependency. Variable 'self' used at line 349 is defined at line 332 and has a Medium-Range dependency. Variable 'buffered_data' used at line 349 is defined at line 332 and has a Medium-Range dependency. Library 'np' used at line 350 is imported at line 19 and has a Long-Range dependency. Variable 'data_sizes' used at line 350 is defined at line 349 and has a Short-Range dependency.

{'List_Comprehension': 1}

{'Variable List_Comprehension Short-Range': 1, 'Variable Medium-Range': 2, 'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ']

[' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)']

['', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 355 is defined at line 332 and has a Medium-Range dependency. Variable 'idxs' used at line 355 is defined at line 352 and has a Short-Range dependency. Variable 'values' used at line 355 is defined at line 353 and has a Short-Range dependency. Variable 'self' used at line 356 is defined at line 332 and has a Medium-Range dependency. Variable 'idxs' used at line 356 is defined at line 352 and has a Short-Range dependency.

{}

{'Variable Medium-Range': 2, 'Variable Short-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """']

[' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)']

[' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 386 is defined at line 378 and has a Short-Range dependency. Variable 'states' used at line 386 is defined at line 378 and has a Short-Range dependency. Library 'tf' used at line 387 is imported at line 20 and has a Long-Range dependency. Variable 'mu' used at line 387 is defined at line 386 and has a Short-Range dependency. Library 'tf' used at line 388 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 388 is defined at line 378 and has a Short-Range dependency. Variable 'mu' used at line 388 is defined at line 387 and has a Short-Range dependency. Library 'tf' used at line 390 is imported at line 20 and has a Long-Range dependency. Variable 'mu' used at line 390 is defined at line 388 and has a Short-Range dependency. Variable 'self' used at line 390 is defined at line 378 and has a Medium-Range dependency. Library 'tf' used at line 392 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 392 is defined at line 378 and has a Medium-Range dependency. Variable 'actions' used at line 392 is defined at line 378 and has a Medium-Range dependency. Library 'tf' used at line 395 is imported at line 20 and has a Long-Range dependency. Variable 'actions' used at line 395 is defined at line 392 and has a Short-Range dependency. Variable 'mu' used at line 395 is defined at line 388 and has a Short-Range dependency. Variable 'std' used at line 395 is defined at line 390 and has a Short-Range dependency. Variable 'self' used at line 395 is defined at line 378 and has a Medium-Range dependency. Library 'tf' used at line 396 is imported at line 20 and has a Long-Range dependency.

{}

{'Variable Short-Range': 9, 'Library Long-Range': 6, 'Variable Medium-Range': 4}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ']

[' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)']

[' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 392 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 392 is defined at line 378 and has a Medium-Range dependency. Variable 'actions' used at line 392 is defined at line 378 and has a Medium-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Medium-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution']

[' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)']

[' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 395 is imported at line 20 and has a Long-Range dependency. Variable 'actions' used at line 395 is defined at line 392 and has a Short-Range dependency. Variable 'mu' used at line 395 is defined at line 388 and has a Short-Range dependency. Variable 'std' used at line 395 is defined at line 390 and has a Short-Range dependency. Variable 'self' used at line 395 is defined at line 378 and has a Medium-Range dependency. Library 'tf' used at line 396 is imported at line 20 and has a Long-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 3, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)']

[' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)']

[' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 410 is imported at line 20 and has a Long-Range dependency. Variable 'mu' used at line 410 is defined at line 409 and has a Short-Range dependency. Library 'tf' used at line 411 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 411 is defined at line 400 and has a Medium-Range dependency. Variable 'mu' used at line 411 is defined at line 410 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 2, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution']

[' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)']

[' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 415}]

Library 'tf' used at line 415 is imported at line 20 and has a Long-Range dependency. Variable 'mu' used at line 415 is defined at line 411 and has a Short-Range dependency. Variable 'self' used at line 415 is defined at line 400 and has a Medium-Range dependency.

{'If Body': 1}

{'Library Long-Range': 1, 'Variable Short-Range': 1, 'Variable Medium-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs']

[' log_probs = self.get_log_prob(states, actions)']

[' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 420 is defined at line 400 and has a Medium-Range dependency. Variable 'states' used at line 420 is defined at line 400 and has a Medium-Range dependency. Variable 'actions' used at line 420 is defined at line 417 and has a Short-Range dependency.

{}

{'Variable Medium-Range': 2, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ']

[' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)']

[' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 423 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 423 is defined at line 400 and has a Medium-Range dependency. Variable 'actions' used at line 423 is defined at line 417 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Medium-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))"]

[" pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)']

[' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 452 is imported at line 20 and has a Long-Range dependency. Variable 'data_rb' used at line 452 is defined at line 447 and has a Short-Range dependency. Variable 'actions_mu' used at line 452 is defined at line 451 and has a Short-Range dependency. Library 'tf' used at line 453 is imported at line 20 and has a Long-Range dependency. Variable 'pi_loss' used at line 453 is defined at line 452 and has a Short-Range dependency.

{}

{'Library Long-Range': 2, 'Variable Short-Range': 3}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)']

[' penalty = orthogonal_regularization(self.actor.base)']

[' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Function 'orthogonal_regularization' used at line 454 is defined at line 122 and has a Long-Range dependency. Variable 'self' used at line 454 is defined at line 447 and has a Short-Range dependency.

{}

{'Function Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ']

[' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)']

[' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'pi_loss_w_penalty' used at line 457 is defined at line 455 and has a Short-Range dependency. Variable 'self' used at line 457 is defined at line 447 and has a Short-Range dependency.

{}

{'Variable Short-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """']

[' skill = tf.zeros((1, self.args.c_dim))']

[' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 501 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 501 is defined at line 497 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill']

[' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))']

[' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Variable 'self' used at line 510 is defined at line 508 and has a Short-Range dependency. Library 'tf' used at line 510 is imported at line 20 and has a Long-Range dependency. Library 'np' used at line 510 is imported at line 19 and has a Long-Range dependency.

{}

{'Variable Short-Range': 1, 'Library Long-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache']

[' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()']

[' ', ' def preprocess_in_state_space(self, item):', ' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)', ' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[{'reason_category': 'If Body', 'usage_line': 562}, {'reason_category': 'If Body', 'usage_line': 563}]

Library 'tf' used at line 562 is imported at line 20 and has a Long-Range dependency. Library 'tf' used at line 563 is imported at line 20 and has a Long-Range dependency.

{'If Body': 2}

{'Library Long-Range': 2}

infilling_python

RL_Motion_Planning

['import os', "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' # Suppress TensorFlow logging", "os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0' # Suppress oneDNN warning", '# os.environ["CUDA_VISIBLE_DEVICES"] = "0"', 'import argparse', 'import datetime', 'import json', 'import os', 'import pickle', 'import sys', 'import time', 'import pickle', 'from abc import ABC', 'from argparse import Namespace', 'from collections import OrderedDict', 'from typing import Dict, Union, List', 'from typing import Tuple', 'import logging', 'import numpy as np', 'import tensorflow as tf', 'import tensorflow_probability as tfp', 'import wandb', 'from keras.layers import Dense', 'from tf_agents.replay_buffers.table import Table', 'from tqdm import tqdm', 'import random', '', '# Set the seed', 'SEED = 1234', '', 'random.seed(SEED)', 'np.random.seed(SEED)', 'tf.random.set_seed(SEED)', 'current_time = datetime.datetime(2024, 1, 1, 0, 0, 0).strftime("%Y%m%d-%H%M%S")', '', "# Ensure TensorFlow doesn't try to use GPU if it's not available", "os.environ['CUDA_VISIBLE_DEVICES'] = '-1' if tf.test.is_gpu_available() else ''", '', '# Suppress other warnings', "tf.get_logger().setLevel('ERROR')", '', '# Get the absolute path of the script module', 'script_path = os.path.realpath(__file__)', 'script_dir = os.path.dirname(script_path)', '', "log_dir = os.path.join(script_dir, './logging', 'BC', 'run' + current_time)", 'if not os.path.exists(log_dir):', ' os.makedirs(log_dir, exist_ok=True)', '', "logging.basicConfig(filename=os.path.join(log_dir, 'logs.txt'), filemode='w',", " format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',", " datefmt='%m/%d/%Y %H:%M:%S',", ' level=logging.INFO)', 'logger = logging.getLogger(__name__)', '', '', 'def get_buffer_shape(args) -> Dict[str, Tuple[int, ...]]:', ' buffer_shape = {', " 'prev_goals': (args.horizon, args.ag_dim),", " 'prev_skills': (args.horizon, args.c_dim),", " 'states': (args.horizon + 1, args.s_dim),", " 'env_goals': (args.horizon + 1, args.g_dim),", " 'curr_goals': (args.horizon, args.ag_dim),", " 'curr_skills': (args.horizon, args.c_dim),", " 'states_2': (args.horizon, args.s_dim),", " 'actions': (args.horizon, args.a_dim),", " 'successes': (args.horizon,),", " 'distances': (args.horizon,),", " 'has_gt_skill': (args.horizon,),", ' }', ' return buffer_shape', '', '', 'def state_to_goal(num_objs: int):', ' """', ' Converts state to goal. (Achieved Goal Space)', ' If obj_identifiers is not None, then it further filters the achieved goals based on the object/skill id.', ' """', ' ', ' @tf.function(experimental_relax_shapes=True) # Imp otherwise code will be very slow', ' def get_goal(states: tf.Tensor, obj_identifiers: tf.Tensor = None):', ' # Get achieved goals', ' goals = tf.map_fn(lambda x: x[3: 3 + num_objs * 3], states, fn_output_signature=tf.float32)', ' return goals', ' ', ' return get_goal', '', '', 'def repurpose_skill_seq(args, skill_seq):', ' """', ' Repurpose the skill sequence to be used for training the policy. Use value of wrap_skill_id', ' = "0": no change', ' = "1": wrap pick/grab/drop:obj_id to pick/grab/drop', ' = "2": wrap pick:obj_id to pick/grab/drop:obj_id to obj_id', ' :param skill_seq: one-hot skill sequence of shape (n_trajs, horizon, c_dim)', ' :return: tensor of shape (n_trajs, horizon, c_dim) and type same as skill_seq', ' """', " if args.env_name != 'OpenAIPickandPlace':", ' tf.print("Wrapping skill sequence is currently only supported for PnP tasks!")', ' sys.exit(-1)', ' ', ' if args.wrap_level == "0":', ' return skill_seq', ' elif args.wrap_level == "1":', ' # wrap by i = j % 3 where i is the new position of skill originally at j. Dim changes from c_dim to 3', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq % 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=3)', ' return skill_seq', ' elif args.wrap_level == "2":', ' # wrap such that 0/1/2 -> 0, 3/4/5 -> 1, 6/7/8 -> 2 ... Dim changes from c_dim to self.args.num_objs', ' skill_seq = tf.argmax(skill_seq, axis=-1)', ' skill_seq = skill_seq // 3', ' # Convert back to one-hot', ' skill_seq = tf.one_hot(skill_seq, depth=args.num_objs)', ' return skill_seq', ' else:', ' raise NotImplementedError("Invalid value for wrap_skill_id: {}".format(args.wrap_level))', '', '', 'def orthogonal_regularization(model, reg_coef=1e-4):', ' """Orthogonal regularization v2.', ' See equation (3) in https://arxiv.org/abs/1809.11096.', ' Rβ(W) = β∥W⊤W ⊙ (1 − I)∥2F, where ⊙ is the Hadamard product.', ' Args:', ' model: A keras model to apply regularization for.', " reg_coef: Orthogonal regularization coefficient. Don't change this value.", ' Returns:', ' A regularization loss term.', ' """', ' reg = 0', ' for layer in model.layers:', ' if isinstance(layer, tf.keras.layers.Dense):', ' prod = tf.matmul(tf.transpose(layer.kernel), layer.kernel)', ' reg += tf.reduce_sum(tf.math.square(prod * (1 - tf.eye(prod.shape[0]))))', ' ', ' print("Orthogonal Regularization: {}".format(reg * reg_coef))', ' return reg * reg_coef', '', '', 'def sample_transitions(sample_style: str, state_to_goal=None, num_options: int = None):', ' def sample_random_transitions(episodic_data, batch_size_in_transitions=None):', ' """', ' Sample random transitions without HER.', ' Functionality: Sample random time-steps from each episode: (g_t-1, c_t-1, s_t, g_t, c_t, a_t) for all episodes.', ' """', ' ', ' batch_size = batch_size_in_transitions # Number of transitions to sample', " T = episodic_data['actions'].shape[1]", " successes = episodic_data['successes']", ' ', ' # Get index at which episode terminated', ' terminate_idxes = tf.math.argmax(successes, axis=-1)', ' # If no success, set to last index', ' mask_no_success = tf.math.equal(terminate_idxes, 0)', ' terminate_idxes += tf.multiply((T - 1) * tf.ones_like(terminate_idxes),', ' tf.cast(mask_no_success, terminate_idxes.dtype))', ' ', " # Get episode idx for each transition to sample: more likely to sample from episodes which didn't end in success", ' p = (terminate_idxes + 1) / tf.reduce_sum(terminate_idxes + 1)', ' episode_idxs = tfp.distributions.Categorical(probs=p).sample(sample_shape=(batch_size,))', ' episode_idxs = tf.cast(episode_idxs, dtype=terminate_idxes.dtype)', ' # Get terminate index for the selected episodes', ' terminate_idxes = tf.gather(terminate_idxes, episode_idxs)', ' print("terminate_idxes: ", terminate_idxes)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # --------------------------------- 2) Select which time steps + goals to use --------------------------------', ' # Get the current time step', ' t_samples_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' t_samples = t_samples_frac * tf.cast(terminate_idxes, dtype=t_samples_frac.dtype)', ' t_samples = tf.cast(tf.round(t_samples), dtype=terminate_idxes.dtype)', ' ', ' # Get random init time step (before t_samples)', ' rdm_past_offset_frac = tf.zeros_like(t_samples_frac)', ' t_samples_init = rdm_past_offset_frac * tf.cast(t_samples, dtype=rdm_past_offset_frac.dtype)', ' t_samples_init = tf.cast(tf.floor(t_samples_init), dtype=t_samples.dtype)', ' print("t_samples_init: ", t_samples_init)', ' ', ' # Get the future time step', ' rdm_future_offset_frac = tf.experimental.numpy.random.random(size=(batch_size,))', ' future_offset = rdm_future_offset_frac * tf.cast((terminate_idxes - t_samples), rdm_future_offset_frac.dtype)', ' future_offset = tf.cast(future_offset, terminate_idxes.dtype)', ' t_samples_future = t_samples + future_offset', ' print("t_samples_future: ", t_samples_future)', ' ', ' # ------------------------------------------------------------------------------------------------------------', ' # ----------------- 3) Select the batch of transitions corresponding to the current time steps ---------------', ' curr_indices = tf.stack((episode_idxs, t_samples), axis=-1)', ' transitions = {}', ' for key in episodic_data.keys():', ' transitions[key] = tf.gather_nd(episodic_data[key], indices=curr_indices)', ' ', " transitions['achieved_goals'] = state_to_goal(", " states=tf.gather_nd(episodic_data['states'], indices=curr_indices),", ' obj_identifiers=None)', ' ', ' # --------------- 4) Select the batch of transitions corresponding to the future time steps ------------', ' future_indices = tf.stack((episode_idxs, t_samples_future), axis=-1)', " transitions['her_goals'] = state_to_goal(states=tf.gather_nd(episodic_data['states'], indices=future_indices),", ' obj_identifiers=None) # Object ids are not used for unsegmented HER', ' ', ' # --------------- 5) Select the batch of transitions corresponding to the initial time steps ------------', ' init_indices = tf.stack((episode_idxs, t_samples_init), axis=-1)', " transitions['init_states'] = tf.gather_nd(episodic_data['states'], indices=init_indices)", ' print("transitions: ", transitions)', ' return transitions', ' ', " if sample_style == 'random_unsegmented':", ' return sample_random_transitions', ' else:', ' raise NotImplementedError', '', '', 'class ReplayBufferTf:', ' def __init__(self, buffer_shapes: Dict[str, Tuple[int, ...]], size_in_transitions, T, transition_fn=None):', ' """Creates a replay buffer.', '', ' Args:', ' buffer_shapes (dict of ints): the shape for all buffers that are used in the replay', ' buffer', ' size_in_transitions (int): the size of the buffer, measured in transitions', ' T (int): the time horizon for episodes', ' transition_fn (function): a function that samples from the replay buffer', ' """', ' self.T = tf.constant(T, dtype=tf.int32)', ' self.buffer_size = tf.constant(size_in_transitions // T, dtype=tf.int32)', ' ', ' self.current_size = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of episodes', ' self.n_transitions_stored = tf.Variable(0, dtype=tf.int32) # Size of buffer in terms of no. of transitions', ' ', ' self.transition_fn = transition_fn', ' self.buffer_keys: List[str] = [key for key in buffer_shapes.keys()]', ' tensor_spec = [tf.TensorSpec(buffer_shapes[key], tf.float32, key) for key in self.buffer_keys]', ' self.table = Table(tensor_spec, capacity=self.buffer_size)', ' ', ' @tf.function # Make sure batch_size passed here is a tf.constant to avoid retracing', ' def sample_transitions(self, batch_size):', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', ' transitions = self.transition_fn(buffered_data, batch_size)', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def sample_episodes(self, ep_start: int = None, ep_end: int = None, num_episodes: int = None):', ' ', ' if ep_start is None or ep_end is None:', ' if num_episodes:', ' num_episodes = tf.math.minimum(tf.cast(num_episodes, dtype=self.current_size.dtype), self.current_size)', ' else:', ' num_episodes = self.current_size', ' ep_range = tf.range(num_episodes)', ' else:', ' ep_range = tf.range(ep_start, ep_end)', ' ', ' buffered_data = {}', ' _data = self.table.read(rows=ep_range)', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' print("buffered_data: ", buffered_data)', ' return buffered_data', ' ', ' @tf.function', ' def store_episode(self, episode_batch):', ' """', ' Store each episode into replay buffer', ' episode_batch: {"": array(1 x (T or T+1) x dim)}', ' """', ' idxs = self._get_storage_idxs(num_to_ins=tf.constant(1, dtype=tf.int32))', ' values = [episode_batch[key] for key in self.buffer_keys if key in episode_batch.keys()]', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + self.T)', ' ', ' def store_episodes(self, episodes_batch):', " for ep_idx in tf.range(tf.shape(episodes_batch['actions'])[0]):", ' episode_batch = {}', ' for key in self.buffer_keys:', ' episode_batch[key] = tf.gather(episodes_batch[key], ep_idx)', ' self.store_episode(episode_batch)', ' ', ' def _get_storage_idxs(self, num_to_ins=None):', ' if num_to_ins is None:', ' num_to_ins = tf.cast(1, dtype=tf.int32)', ' ', ' # consecutively insert until you hit the end of the buffer, and then insert randomly.', ' if self.current_size + num_to_ins <= self.buffer_size:', ' idxs = tf.range(self.current_size, self.current_size + num_to_ins)', ' elif self.current_size < self.buffer_size:', ' overflow = num_to_ins - (self.buffer_size - self.current_size)', ' idx_a = tf.range(self.current_size, self.buffer_size)', ' idx_b = tf.experimental.numpy.random.randint(0, self.current_size, size=(overflow,), dtype=tf.int32)', ' idxs = tf.concat([idx_a, idx_b], axis=0)', ' else:', ' idxs = tf.experimental.numpy.random.randint(0, self.buffer_size, size=(num_to_ins,), dtype=tf.int32)', ' ', ' # update buffer size', ' self.current_size.assign(tf.math.minimum(self.buffer_size, self.current_size + num_to_ins))', ' print("idxs: ", idxs)', ' return idxs', ' ', ' def get_current_size_ep(self):', ' return self.current_size', ' ', ' def get_current_size_trans(self):', ' return self.current_size * self.T', ' ', ' def clear_buffer(self):', ' self.current_size.assign(0)', ' ', ' @property', ' def full(self):', ' return self.current_size == self.buffer_size', ' ', ' def __len__(self):', ' return self.current_size', ' ', ' def save_buffer_data(self, path):', ' buffered_data = {}', ' _data = self.table.read(rows=tf.range(self.current_size))', ' for index, key in enumerate(self.buffer_keys):', ' buffered_data[key] = _data[index]', ' ', " with open(path, 'wb') as handle:", ' pickle.dump(buffered_data, handle, protocol=pickle.HIGHEST_PROTOCOL)', ' ', ' def load_data_into_buffer(self, buffered_data=None, clear_buffer=True, num_demos_to_load=None):', ' ', ' if buffered_data is None:', ' raise ValueError("No buffered_data provided")', ' ', ' if clear_buffer:', ' self.clear_buffer()', ' ', ' if num_demos_to_load is not None:', ' ', ' # Randomly sample idxs to load', " idxs = np.random.choice(len(buffered_data['actions']), size=num_demos_to_load, replace=False).tolist()", ' ', ' for key in buffered_data.keys():', ' buffered_data[key] = tf.gather(buffered_data[key], idxs)', ' ', ' # Check if all tensors are present in loaded data', ' data_sizes = [len(buffered_data[key]) for key in self.buffer_keys]', ' assert np.all(np.array(data_sizes) == data_sizes[0])', ' ', ' idxs = self._get_storage_idxs(num_to_ins=data_sizes[0])', ' values = [buffered_data[key] for key in self.buffer_keys]', ' ', ' self.table.write(rows=idxs, values=values)', ' self.n_transitions_stored.assign(self.n_transitions_stored + len(idxs) * self.T)', '', '', 'class Actor(tf.keras.Model):', ' def __init__(self, action_dim):', ' super(Actor, self).__init__()', ' ', ' # Rewrite the base weights to initialise using Xavier(gain=1.0) and bias=0.0', ' self.base = tf.keras.Sequential([', " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=256, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=128, activation=tf.nn.relu, kernel_initializer='glorot_uniform', bias_initializer='zeros'),", " Dense(units=action_dim, kernel_initializer='glorot_uniform', bias_initializer='zeros')", ' ])', ' ', ' self.MEAN_MIN, self.MEAN_MAX = -7, 7', ' self.eps = np.finfo(np.float32).eps', ' self.pi = tf.constant(np.pi)', ' self.FIXED_STD = 0.05', ' ', ' self.train = True', ' ', ' def get_log_prob(self, states, actions):', ' """Evaluate log probs for actions conditioned on states.', ' Args:', ' states: A batch of states.', ' actions: A batch of actions to evaluate log probs on.', ' Returns:', ' Log probabilities of actions.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' std = tf.ones_like(mu) * self.FIXED_STD', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' ', ' # Get log probs from Gaussian distribution', ' log_probs = -0.5 * tf.square((actions - mu) / std) - 0.5 * tf.math.log(2 * self.pi) - tf.math.log(std)', ' log_probs = tf.reduce_sum(log_probs, axis=1, keepdims=False)', ' print("log_probs: ", log_probs)', ' return log_probs', ' ', ' def call(self, states, training=None, mask=None):', ' """Computes actions for given inputs.', ' Args:', ' states: A batch of states.', ' training: Ignored', ' mask: Ignored.', ' Returns:', ' A mode action, a sampled action and log probability of the sampled action.', ' """', ' mu = self.base(states)', ' mu = tf.nn.tanh(mu)', ' mu = tf.clip_by_value(mu, self.MEAN_MIN, self.MEAN_MAX)', ' ', ' if self.train:', ' # Sample actions from the distribution', ' actions = tf.random.normal(shape=mu.shape, mean=mu, stddev=self.FIXED_STD)', ' else:', ' actions = mu', ' ', ' # Compute log probs', ' log_probs = self.get_log_prob(states, actions)', ' log_probs = tf.expand_dims(log_probs, -1) # To avoid broadcasting', ' ', ' actions = tf.clip_by_value(actions, -1 + self.eps, 1 - self.eps)', ' print("mu: ", mu)', ' print("actions: ", actions)', ' print("log_probs: ", log_probs)', ' return mu, actions, log_probs', '', '', 'class BC(tf.keras.Model, ABC):', ' def __init__(self, args: Namespace):', ' super(BC, self).__init__()', ' self.args = args', ' ', ' # Declare Policy Network and Optimiser', ' self.actor = Actor(args.a_dim)', ' self.actor_optimizer = tf.keras.optimizers.Adam(learning_rate=args.actor_lr)', ' ', ' # Build Model', ' self.build_model()', ' ', ' # For HER', ' self.use_her = False', " logger.info('[[[ Using HER ? ]]]: {}'.format(self.use_her))", ' ', ' @tf.function(experimental_relax_shapes=True)', ' def train(self, data_exp, data_rb):', ' with tf.GradientTape(watch_accessed_variables=False, persistent=True) as tape:', ' tape.watch(self.actor.variables)', ' ', " actions_mu, _, _ = self.actor(tf.concat([data_rb['states'], data_rb['goals']], axis=1))", " pi_loss = tf.reduce_sum(tf.math.squared_difference(data_rb['actions'], actions_mu), axis=-1)", ' pi_loss = tf.reduce_mean(pi_loss)', ' penalty = orthogonal_regularization(self.actor.base)', ' pi_loss_w_penalty = pi_loss + penalty', ' ', ' grads = tape.gradient(pi_loss_w_penalty, self.actor.trainable_variables)', ' self.actor_optimizer.apply_gradients(zip(grads, self.actor.trainable_variables))', ' print("loss/pi: ", pi_loss)', ' print("penalty/pi_ortho_penalty: ", penalty)', ' return {', " 'loss/pi': pi_loss,", " 'penalty/pi_ortho_penalty': penalty,", ' }', ' ', ' def act(self, state, env_goal, prev_goal, prev_skill, epsilon, stddev):', ' state = tf.clip_by_value(state, -self.args.clip_obs, self.args.clip_obs)', ' env_goal = tf.clip_by_value(env_goal, -self.args.clip_obs, self.args.clip_obs)', ' prev_goal = tf.clip_by_value(prev_goal, -self.args.clip_obs, self.args.clip_obs)', ' ', ' # ###################################### Current Goal ####################################### #', ' curr_goal = env_goal', ' ', ' # ###################################### Current Skill ###################################### #', ' curr_skill = prev_skill # Not used in this implementation', ' ', ' # ########################################## Action ######################################### #', ' # Explore', ' if tf.random.uniform(()) < epsilon:', ' action = tf.random.uniform((1, self.args.a_dim), -self.args.action_max, self.args.action_max)', ' # Exploit', ' else:', ' action_mu, _, _ = self.actor(tf.concat([state, curr_goal], axis=1)) # a_t = mu(s_t, g_t)', ' action_dev = tf.random.normal(action_mu.shape, mean=0.0, stddev=stddev)', ' action = action_mu + action_dev # Add noise to action', ' action = tf.clip_by_value(action, -self.args.action_max, self.args.action_max)', ' ', ' # Safety check for action, should not be nan or inf', ' has_nan = tf.math.reduce_any(tf.math.is_nan(action))', ' has_inf = tf.math.reduce_any(tf.math.is_inf(action))', ' if has_nan or has_inf:', " logger.warning('Action has nan or inf. Setting action to zero. Action: {}'.format(action))", ' action = tf.zeros_like(action)', ' ', ' return curr_goal, curr_skill, action', ' ', ' def get_init_skill(self):', ' """', ' demoDICE does not use skills. Use this function to return a dummy skill of dimension (1, c_dim)', ' """', ' skill = tf.zeros((1, self.args.c_dim))', ' return skill', ' ', ' @staticmethod', ' def get_init_goal(init_state, g_env):', ' return g_env', ' ', ' def build_model(self):', ' # a_t <- f(s_t) for each skill', ' _ = self.actor(tf.concat([np.ones([1, self.args.s_dim]), np.ones([1, self.args.g_dim])], 1))', ' ', ' def save_(self, dir_param):', ' self.actor.save_weights(dir_param + "/policy.h5")', ' ', ' def load_(self, dir_param):', ' self.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def change_training_mode(self, training_mode: bool):', ' pass', ' ', ' def update_target_networks(self):', ' pass', '', '', 'class AgentBase(object):', ' def __init__(', ' self,', ' args,', ' model,', ' algo: str,', ' expert_buffer: ReplayBufferTf,', ' offline_buffer: ReplayBufferTf', ' ):', ' ', ' self.args = args', ' self.model = model', ' ', ' # Define the Buffers', ' self.expert_buffer = expert_buffer', ' self.offline_buffer = offline_buffer', ' ', ' self.offline_gt_prev_skill = None', ' self.offline_gt_curr_skill = None', ' ', ' # Define Tensorboard for logging Losses and Other Metrics', ' if not os.path.exists(args.dir_summary):', ' os.makedirs(args.dir_summary)', ' ', ' if not os.path.exists(args.dir_plot):', ' os.makedirs(args.dir_plot)', ' self.summary_writer = tf.summary.create_file_writer(args.dir_summary)', ' ', ' # Define wandb logging', ' if self.args.log_wandb:', ' self.wandb_logger = wandb.init(', ' project=args.wandb_project,', ' config=vars(args),', " id='{}_{}'.format(algo, current_time),", ' reinit=True, # Allow multiple wandb.init() calls in the same process.', ' )', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' def preprocess_in_state_space(self, item):']

[' item = tf.clip_by_value(item, -self.args.clip_obs, self.args.clip_obs)']

[' return item', ' ', ' def save_model(self, dir_param):', ' if not os.path.exists(dir_param):', ' os.makedirs(dir_param)', ' self.model.save_(dir_param)', ' ', ' def load_model(self, dir_param):', ' self.model.load_(dir_param)', ' ', ' def process_data(self, transitions, expert=False, is_supervised=False):', ' ', ' trans = transitions.copy()', ' ', ' # Process the states and goals', " trans['states'] = self.preprocess_in_state_space(trans['states'])", " trans['states_2'] = self.preprocess_in_state_space(trans['states_2'])", " trans['env_goals'] = self.preprocess_in_state_space(trans['env_goals'])", " trans['init_states'] = self.preprocess_in_state_space(trans['init_states'])", " trans['her_goals'] = self.preprocess_in_state_space(trans['her_goals'])", " trans['achieved_goals'] = self.preprocess_in_state_space(trans['achieved_goals'])", ' ', ' if self.model.use_her:', " trans['goals'] = trans['her_goals']", ' else:', " trans['goals'] = trans['env_goals']", ' ', ' # Define if the transitions are from expert or not/are supervised or not', " trans['is_demo'] = tf.cast(expert, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", " trans['is_sup'] = tf.cast(is_supervised, dtype=tf.int32) * tf.ones_like(trans['successes'], dtype=tf.int32)", ' ', ' # Compute terminate skills i.e. if prev_skill != curr_skill then terminate_skill = 1 else 0', " trans['terminate_skills'] = tf.cast(tf.not_equal(tf.argmax(trans['prev_skills'], axis=-1),", " tf.argmax(trans['curr_skills'], axis=-1)),", ' dtype=tf.int32)', ' # reshape the terminate_skills to be of shape (batch_size, 1)', " trans['terminate_skills'] = tf.reshape(trans['terminate_skills'], shape=(-1, 1))", ' ', ' # Make sure the data is of type tf.float32', ' for key in trans.keys():', ' trans[key] = tf.cast(trans[key], dtype=tf.float32)', ' print("trans :", trans)', ' return trans', ' ', ' def sample_data(self, buffer, batch_size):', ' ', ' # Sample Transitions', ' transitions: Union[Dict[int, dict], dict] = buffer.sample_transitions(batch_size)', ' ', ' # Process the transitions', ' keys = None', ' if all(isinstance(v, dict) for v in transitions.values()):', ' for skill in transitions.keys():', ' ', ' # For skills whose transition data is not None', ' if transitions[skill] is not None:', ' transitions[skill] = self.process_data(', ' transitions[skill], tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' keys = transitions[skill].keys()', ' ', ' # If keys is None, No transitions were sampled', ' if keys is None:', ' raise ValueError("No transitions were sampled")', ' ', ' # Concatenate the transitions from different skills', ' combined_transitions = {key: [] for key in keys}', ' ', ' for skill in transitions.keys():', ' ', ' if transitions[skill] is not None:', ' for key in keys:', ' combined_transitions[key].append(transitions[skill][key])', ' ', ' for key in keys:', ' combined_transitions[key] = tf.concat(combined_transitions[key], axis=0)', ' ', ' transitions = combined_transitions', ' ', ' elif isinstance(transitions, dict):', ' transitions = self.process_data(', ' transitions, tf.constant(True, dtype=tf.bool), tf.constant(True, dtype=tf.bool)', ' )', ' ', ' else:', ' raise ValueError("Invalid type of transitions")', ' print("transitions: ", transitions)', ' return transitions', ' ', ' @tf.function', ' def train(self):', ' ', ' self.model.change_training_mode(training_mode=True)', ' ', ' data_expert = self.sample_data(self.expert_buffer, self.args.batch_size)', ' data_policy = self.sample_data(self.offline_buffer, self.args.batch_size)', ' loss_dict = self.model.train(data_expert, data_policy)', ' ', ' # Average the losses', ' avg_loss_dict = {}', ' for key in loss_dict.keys():', ' if key not in avg_loss_dict.keys():', ' avg_loss_dict[key] = []', ' avg_loss_dict[key].append(loss_dict[key])', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = tf.reduce_mean(avg_loss_dict[key])', ' print("avg_loss_dict: ", avg_loss_dict)', ' return avg_loss_dict', ' ', ' def learn(self):', ' # This is a base class method, inherited classes must implement this method', ' raise NotImplementedError', '', '', 'class Agent(AgentBase):', ' def __init__(self, args,', ' expert_buffer: ReplayBufferTf = None,', ' offline_buffer: ReplayBufferTf = None):', ' ', " super(Agent, self).__init__(args, BC(args), 'BC', expert_buffer, offline_buffer)", ' ', ' def load_actor(self, dir_param):', ' self.model.actor.load_weights(dir_param + "/policy.h5")', ' ', ' def learn(self):', ' args = self.args', ' ', ' # Tracker for wandb logging', ' log_step = 0', ' ', ' # [Update] Load the expert data into the expert buffer, expert data and offline data into the offline buffer', ' data_exp = self.expert_buffer.sample_episodes()', ' data_off = self.offline_buffer.sample_episodes()', ' self.expert_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_exp, clear_buffer=True)', ' self.offline_buffer.load_data_into_buffer(buffered_data=data_off, clear_buffer=False)', ' ', ' with tqdm(total=args.max_time_steps, leave=False) as pbar:', ' for curr_t in range(0, args.max_time_steps):', ' ', ' # Update the reference actors and directors using polyak averaging', ' if curr_t % args.update_target_interval == 0:', ' tf.print("Updating the target actors and critics at train step {}".format(curr_t))', ' self.model.update_target_networks()', ' ', ' # Train the policy', " pbar.set_description('Training')", ' avg_loss_dict = self.train()', ' for key in avg_loss_dict.keys():', ' avg_loss_dict[key] = avg_loss_dict[key].numpy().item()', ' ', ' # Log', ' if self.args.log_wandb:', ' self.wandb_logger.log(avg_loss_dict, step=log_step)', ' self.wandb_logger.log({', " 'policy_buffer_size': self.offline_buffer.get_current_size_trans(),", " 'expert_buffer_size': self.expert_buffer.get_current_size_trans(),", ' }, step=log_step)', ' ', ' # Update', ' pbar.update(1)', ' log_step += 1', ' ', ' # Save the model', ' self.save_model(args.dir_param)', '', '', 'def get_config_env(args, ag_in_env_goal):', ' """', ' :param args: Namespace object', ' :param ag_in_env_goal: If True, then achieved goal is in the same space as env goal', ' """', ' ', ' args.g_dim = 3', ' args.s_dim = 10', ' args.a_dim = 4', ' ', " # Specify the expert's latent skill dimension [Default]", " # Define number of skills, this could be different from agent's practiced skill dimension", " assert hasattr(args, 'num_objs')", ' args.c_dim = 3 * args.num_objs', ' ', ' if ag_in_env_goal:', ' args.ag_dim = args.g_dim # Achieved Goal in the same space as Env Goal', ' else:', ' args.ag_dim = 3 # Goal/Object position in the 3D space', ' print("args: ", args)', ' return args', '', '', 'def get_config(db=False):', ' # Construct the absolute path of the data directory', " data_dir = os.path.join(script_dir, 'pnp_data')", '', ' parser = argparse.ArgumentParser()', ' ', " parser.add_argument('--log_wandb', type=bool, default=False)", " parser.add_argument('--wandb_project', type=str, default='offlineILPnPOne',", " choices=['offlineILPnPOne', 'offlineILPnPOneExp', 'offlineILPnPTwoExp'])", ' ', " parser.add_argument('--expert_demos', type=int, default=25)", " parser.add_argument('--offline_demos', type=int, default=75)", " parser.add_argument('--eval_demos', type=int, default=1 if db else 10,", " help='Use 10 (num of demos to evaluate trained pol)')", " parser.add_argument('--test_demos', type=int, default=0, help='For Visualisation')", " parser.add_argument('--perc_train', type=int, default=1.0)", ' ', ' # Specify Environment Configuration', " parser.add_argument('--env_name', type=str, default='OpenAIPickandPlace')", " parser.add_argument('--num_objs', type=int, default=1)", " parser.add_argument('--horizon', type=int, default=100,", " help='Set 100 for one_obj, 150 for two_obj and 200 for three_obj')", " parser.add_argument('--stacking', type=bool, default=False)", " parser.add_argument('--expert_behaviour', type=str, default='0', choices=['0', '1'],", " help='Expert behaviour in two_object env')", " parser.add_argument('--full_space_as_goal', type=bool, default=False)", " parser.add_argument('--fix_goal', type=bool, default=False,", " help='[Debugging] Fix the goal position for one object task')", " parser.add_argument('--fix_object', type=bool, default=False,", " help='[Debugging] Fix the object position for one object task')", ' ', ' # Specify Data Collection Configuration', " parser.add_argument('--buffer_size', type=int, default=int(2e5),", " help='Number of transitions to store in buffer (max_time_steps)')", ' ', ' # Specify Training configuration', " parser.add_argument('--max_pretrain_time_steps', type=int, default=0 if not db else 0,", " help='No. of time steps to run pretraining - actor, director on expert data. Set to 0 to skip')", " parser.add_argument('--max_time_steps', type=int, default=10000 if not db else 1,", " help='No. of time steps to run. Recommended 5k for one_obj, 10k for two_obj')", " parser.add_argument('--batch_size', type=int, default=1,", " help='No. of trans to sample from buffer for each update')", " parser.add_argument('--trans_style', type=str, default='random_unsegmented',", " choices=['random_unsegmented', 'random_segmented'],", " help='How to sample transitions from expert buffer')", ' ', ' # Viterbi configuration', " parser.add_argument('--skill_supervision', type=str, default='none',", " choices=['full', 'semi:0.10', 'semi:0.25', 'none'],", " help='Type of supervision for latent skills. '", " 'full: Use ground truth skills for offline data.'", " 'semi:x: Use Viterbi to update latent skills for offline data.'", " 'none: Use Viterbi to update latent skills for expert and offline data.')", " parser.add_argument('--num_skills', type=int, default=None,", " help='Number of skills to use for agent, if provided, will override expert skill set. '", ' \'Use when skill supervision is "none"\')', " parser.add_argument('--wrap_level', type=str, default='1', choices=['0', '1', '2'],", " help='consumed by multi-object expert to determine how to wrap effective skills of expert')", ' ', ' # Polyak', " parser.add_argument('--update_target_interval', type=int, default=20,", " help='Number of time steps after which target networks will be updated using polyak averaging')", " parser.add_argument('--actor_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for actor.')", " parser.add_argument('--director_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for director.')", " parser.add_argument('--critic_polyak', type=float, default=0.95,", " help='Polyak averaging coefficient for critic.')", ' ', ' # Evaluation', " parser.add_argument('--eval_interval', type=int, default=100)", " parser.add_argument('--visualise_test', type=bool, default=False, help='Visualise test episodes?')", ' ', ' # Parameters', " parser.add_argument('--discount', type=float, default=0.99, help='Discount used for returns.')", " parser.add_argument('--replay_regularization', type=float, default=0.05,", " help='Replay Regularization Coefficient. Used by both ValueDICE (0.1) and DemoDICE (0.05)')", " parser.add_argument('--nu_grad_penalty_coeff', type=float, default=1e-4,", " help='Nu Net Gradient Penalty Coefficient. ValueDICE uses 10.0, DemoDICE uses 1e-4')", " parser.add_argument('--cost_grad_penalty_coeff', type=float, default=10,", " help='Cost Net Gradient Penalty Coefficient')", " parser.add_argument('--actor_lr', type=float, default=3e-3)", " parser.add_argument('--critic_lr', type=float, default=3e-4)", " parser.add_argument('--disc_lr', type=float, default=3e-4)", " parser.add_argument('--clip_obs', type=float, default=200.0,", " help='Un-normalised i.e. raw Observed Values (State and Goals) are clipped to this value')", ' ', ' # Specify Path Configurations', " parser.add_argument('--dir_data', type=str, default=data_dir)", " parser.add_argument('--dir_root_log', type=str, default=log_dir)", " parser.add_argument('--dir_summary', type=str, default=os.path.join(log_dir, 'summary'))", " parser.add_argument('--dir_plot', type=str, default=os.path.join(log_dir, 'plots'))", " parser.add_argument('--dir_param', type=str, default=os.path.join(log_dir, 'models'))", " parser.add_argument('--dir_post', type=str, default='./finetuned_models',", " help='Provide the <path_to_models>')", " parser.add_argument('--dir_pre', type=str, default='./pretrained_models',", " help='Provide the <path_to_models>')", ' ', ' args = parser.parse_args()', ' ', ' # Load the environment config', ' args = get_config_env(args, ag_in_env_goal=True)', ' ', ' # Other Configurations', ' args.train_demos = int(args.expert_demos * args.perc_train)', ' args.val_demos = args.expert_demos - args.train_demos', ' ', ' # Set number of skills [For unsupervised skill learning]', " if args.num_skills is not None and args.skill_supervision == 'none':", " print('Overriding c_dim with specified %d skills' % args.num_skills)", ' args.c_dim = args.num_skills', ' ', ' # Set number of skills [For full or semi-supervised skill learning]', " if args.env_name == 'OpenAIPickandPlace' and args.wrap_level != '0' and args.skill_supervision != 'none':", " print('Overriding c_dim based on Wrap Level %s' % args.wrap_level)", " if args.wrap_level == '1':", ' args.c_dim = 3', " elif args.wrap_level == '2':", ' args.c_dim = args.num_objs', ' else:', " raise NotImplementedError('Wrap level %s not implemented' % args.wrap_level)", ' ', ' return args', '', '', 'def run(db: bool, algo: str):', ' ', ' if db:', ' print("Running in Debug Mode. (db=True)")', ' ', ' tf.config.run_functions_eagerly(db)', ' ', ' logger.info("# ################# Working on Model: \\"{}\\" ################# #".format(algo))', ' ', ' args = get_config(db=db)', ' args.algo = algo', ' args.log_dir = log_dir', ' ', ' logger.info("---------------------------------------------------------------------------------------------")', ' config: dict = vars(args)', ' config = {key: str(value) for key, value in config.items()}', ' config = OrderedDict(sorted(config.items()))', ' logger.info(json.dumps(config, indent=4))', ' ', ' # Clear tensorflow graph and cache', ' tf.keras.backend.clear_session()', ' tf.compat.v1.reset_default_graph()', ' ', ' # ######################################################################################################## #', ' # ############################################# DATA LOADING ############################################# #', ' # ######################################################################################################## #', ' # Load Buffer to store expert data', ' n_objs = args.num_objs', ' buffer_shape: Dict[str, Tuple[int, ...]] = get_buffer_shape(args)', ' ', ' expert_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim),', ' )', ' offline_buffer = ReplayBufferTf(', ' buffer_shape, args.buffer_size, args.horizon,', ' sample_transitions(args.trans_style, state_to_goal=state_to_goal(n_objs), num_options=args.c_dim)', ' )', ' if n_objs == 3:', " expert_data_file = 'three_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'three_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 2:', " expert_data_file = 'two_obj_{}_train.pkl'.format(args.expert_behaviour)", " offline_data_file = 'two_obj_{}_offline.pkl'.format(args.expert_behaviour)", ' elif n_objs == 1:', " expert_data_file = 'single_obj_train.pkl'", " offline_data_file = 'single_obj_offline.pkl'", ' else:', ' raise NotImplementedError', ' expert_data_path = os.path.join(args.dir_data, expert_data_file)', ' offline_data_path = os.path.join(args.dir_data, offline_data_file)', ' ', ' if not os.path.exists(expert_data_path):', ' logger.error(', ' "Expert data not found at {}. Please run the data generation script first.".format(expert_data_path))', ' sys.exit(-1)', ' ', ' if not os.path.exists(offline_data_path):', ' logger.error(', ' "Offline data not found at {}. Please run the data generation script first.".format(offline_data_path))', ' sys.exit(-1)', ' ', ' # Store the expert data in the expert buffer -> D_E', ' logger.info("Loading Expert Demos from {} into Expert Buffer for training.".format(expert_data_path))', " with open(expert_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' expert_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.expert_demos)', ' ', ' # Store the offline data in the policy buffer for DemoDICE -> D_O', ' logger.info("Loading Offline Demos from {} into Offline Buffer for training.".format(offline_data_path))', " with open(offline_data_path, 'rb') as handle:", ' buffered_data = pickle.load(handle)', ' ', ' # [Optional] Reformat the G.T. skill sequences', " curr_skills = repurpose_skill_seq(args, buffered_data['curr_skills'])", " prev_skills = repurpose_skill_seq(args, buffered_data['prev_skills'])", " buffered_data['curr_skills'] = curr_skills", " buffered_data['prev_skills'] = prev_skills", ' # Add a new key "has_gt_skill" indicating that the skill is G.T.', " buffered_data['has_gt_skill'] = tf.ones_like(buffered_data['successes'], dtype=tf.float32)", ' offline_buffer.load_data_into_buffer(buffered_data=buffered_data, num_demos_to_load=args.offline_demos)', ' # ########################################################################################################### #', ' # ############################################# TRAINING #################################################### #', ' # ########################################################################################################### #', ' start = time.time()', ' ', ' agent = Agent(args, expert_buffer, offline_buffer)', ' ', ' logger.info("Training .......")', ' agent.learn()', '', '', 'if __name__ == "__main__":', ' num_runs = 1', ' for i in range(num_runs):', " run(db=True, algo='BC')"]

[]

Library 'tf' used at line 566 is imported at line 20 and has a Long-Range dependency. Variable 'self' used at line 566 is defined at line 565 and has a Short-Range dependency.

{}

{'Library Long-Range': 1, 'Variable Short-Range': 1}