first commit

app.py

@@ -1,12 +1,12 @@
+import spaces
 import gradio as gr
 from transformers import pipeline
 import torch
 import os
-import spaces
 
 hf_token = os.environ["HF_TOKEN"]
 
-# Load the Gemma 3 pipeline.
+# Load the Gemma 3 pipeline
 pipe = pipeline(
     "image-text-to-text",
     model="google/gemma-3-4b-it",
@@ -14,45 +14,50 @@ pipe = pipeline(
     torch_dtype=torch.bfloat16,
     use_auth_token=hf_token
 )
+
 @spaces.GPU
 def generate_response(user_text, user_image):
-    # Check if an image was uploaded.
     if user_image is None:
-        return "Error: An image upload is mandatory."
+        return "Please upload an image (required)"
     
-    # Prepare messages with the system prompt and user inputs.
     messages = [
         {
             "role": "system",
             "content": [{"type": "text", "text": "You are a helpful assistant."}]
         }
     ]
+    
     user_content = [{"type": "image", "image": user_image}]
     if user_text:
         user_content.append({"type": "text", "text": user_text})
+    
     messages.append({"role": "user", "content": user_content})
     
-    # Call the pipeline.
+    # Call the pipeline with the provided messages
     output = pipe(text=messages, max_new_tokens=200)
     
-    # Try to extract the generated content.
     try:
         response = output[0]["generated_text"][-1]["content"]
+        return response
     except (KeyError, IndexError, TypeError):
-        response = str(output)
-    
-    return response
+        return "Error processing the response. Please try again."
 
-iface = gr.Interface(
-    fn=generate_response,
-    inputs=[
-        gr.Textbox(label="Message", placeholder="Type your message here..."),
-        gr.Image(type="pil", label="Upload an Image", source="upload")
-    ],
-    outputs=gr.Textbox(label="Response"),
-    title="Gemma 3 Simple Interface",
-    description="Enter your message and upload an image (image upload is mandatory) to get a response."
-)
+with gr.Blocks() as demo:
+    gr.Markdown("# Gemma 3 Image Analysis")
+    gr.Markdown("Upload an image and optionally add a prompt to get the model's response.")
+    
+    with gr.Row():
+        img = gr.Image(type="pil", label="Upload an image (required)")
+        txt = gr.Textbox(label="Your prompt (optional)", placeholder="Describe what you see in this image")
+    
+    output = gr.Textbox(label="Model Response")
+    
+    submit_btn = gr.Button("Submit")
+    submit_btn.click(
+        generate_response, 
+        inputs=[txt, img], 
+        outputs=output
+    )
 
 if __name__ == "__main__":
-    iface.launch()
+    demo.launch()