Update app.py

app.py

@@ -4,11 +4,11 @@ from transformers import pipeline, AutoModelForVision2Seq, AutoProcessor
 import torch
 
 # Load the OpenGVLab/InternVL-Chat-V1-5 model and processor
-processor = AutoProcessor.from_pretrained("OpenGVLab/InternVL-Chat-V1-5")
-model = AutoModelForVision2Seq.from_pretrained("OpenGVLab/InternVL-Chat-V1-5")
+from transformers import AutoModel
+model = AutoModel.from_pretrained("OpenGVLab/InternVL-Chat-V1-5", trust_remote_code=True)
 
 # Load the Llama3 model for text processing
-llama_model = pipeline("text2text-generation", model="llama3")
+#llama_model = pipeline("text2text-generation", model="llama3")
 
 def process_image(image):
     # Process the image to extract the recipe using OpenGVLab
@@ -18,25 +18,17 @@ def process_image(image):
 
     return extracted_text
 
-def adjust_recipe(extracted_text, adjustment):
-    # Create the prompt for Llama3 to adjust the recipe
-    prompt = f"Here is a recipe: {extracted_text}. Please {adjustment} the recipe."
-    response = llama_model(prompt)
-    return response[0]['generated_text']
-
-def app(image, adjustment):
-    extracted_text = process_image(image)
-    adjusted_recipe = adjust_recipe(extracted_text, adjustment)
-    return adjusted_recipe
-
-# Create the Gradio interface
-interface = gr.Interface(
-    fn=app,
-    inputs=[gr.inputs.Image(type="file"), gr.inputs.Dropdown(["double", "halve"])],
+iface = gr.Interface(
+    fn=process_image,
+    inputs=[
+        gr.components.Image(type="filepath", label="Recipe Image"),
+        #gr.components.Radio(choices=["Double", "Triple", "Half", "Third"], label="Action")
+    ],
     outputs="text",
-    title="Recipe Adjuster",
-    description="Upload an image of a recipe, and this app will double or halve the recipe."
+    title="Recipe Modifier",
+    description="Upload an image of a recipe and choose how to modify the measurements.",
 )
 
+
 if __name__ == "__main__":
-    interface.launch()
+    interface.launch(share=True)