Update app.py

app.py

@@ -4,18 +4,9 @@ import json
 import re
 from gradio_client import Client
 
-#fusecap_client = Client("https://noamrot-fusecap-image-captioning.hf.space/")
-#fuyu_client = Client("https://adept-fuyu-8b-demo.hf.space/")
 kosmos2_client = Client("https://ydshieh-kosmos-2.hf.space/")
 
 def get_caption(image_in):
-    """
-    fuyu_result = fuyu_client.predict(
-	    image_in,	# str representing input in 'raw_image' Image component
-	    True,	# bool  in 'Enable detailed captioning' Checkbox component
-		fn_index=2
-    )
-    """
 
     kosmos2_result = kosmos2_client.predict(
         image_in,	# str (filepath or URL to image) in 'Test Image' Image component
@@ -67,13 +58,12 @@ def get_caption_from_MD(image_in):
     return result
 
 def get_magnet(prompt):
-    amended_prompt = f"{prompt}"
-    print(amended_prompt)
+
     client = Client("https://fffiloni-magnet.hf.space/")
     result = client.predict(
         "facebook/magnet-medium-10secs",	# Literal['facebook/magnet-small-10secs', 'facebook/magnet-medium-10secs', 'facebook/magnet-small-30secs', 'facebook/magnet-medium-30secs', 'facebook/audio-magnet-small', 'facebook/audio-magnet-medium']  in 'Model' Radio component
         "",	# str  in 'Model Path (custom models)' Textbox component
-        amended_prompt,	# str  in 'Input Text' Textbox component
+        prompt,	# str  in 'Input Text' Textbox component
         3,	# float  in 'Temperature' Number component
         0.9,	# float  in 'Top-p' Number component
         10,	# float  in 'Max CFG coefficient' Number component
@@ -112,7 +102,7 @@ instruction = f"""
 <|user|>
 """
 
-@spaces.GPU(enable_queue=True, duration=60)
+@spaces.GPU(enable_queue=True)
 def get_musical_prompt(user_prompt):
     prompt = f"{instruction.strip()}\n{user_prompt}</s>"    
     outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
@@ -132,7 +122,7 @@ def infer(image_in):
     gr.Info("Now calling MAGNet for music ...")
     music_o = get_magnet(musical_prompt)
     
-    return cleaned_text, music_o
+    return musical_prompt, music_o
 
 demo_title = "Image to Music V2"
 description = "Get music from a picture"
@@ -168,7 +158,7 @@ with gr.Blocks(css=css) as demo:
                 result = gr.Audio(
                     label = "Music"
                 )
-        with gr.Row():
+        with gr.Column():
             gr.Examples(
                 examples = [
                     ["examples/monalisa.png"],