xtuner
/

llava-v1.5-7b-xtuner

Image-Text-to-Text

Model card Files Files and versions Community

LZHgrla commited on Dec 16, 2023

Commit

5544a13

•

1 Parent(s): dcd50b3

Update xtuner_config.py

Files changed (1) hide show

xtuner_config.py +4 -4

xtuner_config.py CHANGED Viewed

@@ -20,7 +20,7 @@ custom_hooks = [
             'Please describe this picture',
         ],
         every_n_iters=500,
-        processor=dict(
             pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
             trust_remote_code=True,
             type='transformers.CLIPImageProcessor.from_pretrained'),
@@ -60,7 +60,7 @@ llava_dataset = dict(
     image_folder='./data/llava_data/llava_images',
     max_length=1472,
     pad_image_to_square=True,
-    processor=dict(
         pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
         trust_remote_code=True,
         type='transformers.CLIPImageProcessor.from_pretrained'),
@@ -144,7 +144,7 @@ param_scheduler = [
         type='mmengine.optim.CosineAnnealingLR'),
 ]
 pretrained_pth = './work_dirs/llava_vicuna_7b_v15_clip_vit_large_p14_336_e1_gpu8_pretrain/epoch_1.pth'
-processor = dict(
     pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
     trust_remote_code=True,
     type='transformers.CLIPImageProcessor.from_pretrained')
@@ -183,7 +183,7 @@ train_dataloader = dict(
         image_folder='./data/llava_data/llava_images',
         max_length=1472,
         pad_image_to_square=True,
-        processor=dict(
             pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
             trust_remote_code=True,
             type='transformers.CLIPImageProcessor.from_pretrained'),

             'Please describe this picture',
         ],
         every_n_iters=500,
+        image_processor=dict(
             pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
             trust_remote_code=True,
             type='transformers.CLIPImageProcessor.from_pretrained'),
     image_folder='./data/llava_data/llava_images',
     max_length=1472,
     pad_image_to_square=True,
+    image_processor=dict(
         pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
         trust_remote_code=True,
         type='transformers.CLIPImageProcessor.from_pretrained'),
         type='mmengine.optim.CosineAnnealingLR'),
 ]
 pretrained_pth = './work_dirs/llava_vicuna_7b_v15_clip_vit_large_p14_336_e1_gpu8_pretrain/epoch_1.pth'
+image_processor = dict(
     pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
     trust_remote_code=True,
     type='transformers.CLIPImageProcessor.from_pretrained')
         image_folder='./data/llava_data/llava_images',
         max_length=1472,
         pad_image_to_square=True,
+        image_processor=dict(
             pretrained_model_name_or_path='openai/clip-vit-large-patch14-336',
             trust_remote_code=True,
             type='transformers.CLIPImageProcessor.from_pretrained'),