Update app.py

app.py

@@ -32,10 +32,8 @@ def respond(
         if val[1]:
             messages.append({"role": "assistant", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
-
     if image:
-        # Convert image(s) to base64
+        # Convert image to base64
         if isinstance(image, Image.Image):
             image_b64 = image_to_base64(image)
             messages.append({"role": "user", "content": "Image uploaded", "image": image_b64})
@@ -44,9 +42,10 @@ def respond(
                 image_b64 = image_to_base64(img)
                 messages.append({"role": "user", "content": "Image uploaded", "image": image_b64})
 
+    messages.append({"role": "user", "content": message})
+
     try:
         responses = []
-        generated_image = None
 
         for response in client.chat_completion(
             messages,
@@ -58,18 +57,10 @@ def respond(
             token = response.choices[0].delta.content
             responses.append(token)
 
-            # Check if the response contains an image to be displayed
-            if response.choices[0].delta.image:
-                image_b64 = response.choices[0].delta.image
-                image_data = base64.b64decode(image_b64)
-                generated_image = Image.open(io.BytesIO(image_data))
-                # Optionally convert to RGB if needed
-                # generated_image = generated_image.convert("RGB")
-
-        return responses, generated_image
+        return responses
 
     except Exception as e:
-        return [str(e)], None
+        return [str(e)]
 
 # Debugging print statements
 print("Starting Gradio interface setup...")
@@ -78,15 +69,12 @@ try:
     demo = gr.Interface(
         fn=respond,
         inputs=[
-            gr.Textbox(label="Message"),
-            gr.Image(label="Upload Medical Image (Optional)", type="pil")
-        ],
-        outputs=[
-            gr.Textbox(label="Response", placeholder="Model response will appear here..."),
-            gr.Image(label="Generated Image", type="pil", output=True)
+            gr.Image(label="Upload Medical Image", type="pil", optional=True),
+            gr.Textbox(label="Message")
         ],
-        title="LLAVA Model - Medical Image and Text",
-        description="Upload a medical image and ask a specific question about the image or provide text input for a medical description.",
+        outputs=gr.Textbox(label="Response", placeholder="Model response will appear here..."),
+        title="LLAVA Model - Medical Image and Question",
+        description="Upload a medical image and ask a specific question about the image for a medical description.",
         additional_inputs=[
             gr.Textbox(label="System message", value="You are a friendly Chatbot."),
             gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),