Update app.py

app.py

@@ -33,10 +33,20 @@ def predict(image, text):
     inputs = processor(image, input_text, return_tensors="pt").to(model.device)
     outputs = model.generate(**inputs, max_new_tokens=250)
     response = processor.decode(outputs[0], skip_special_tokens=True)
-    # Split the response at the first occurrence of "assistant" and return only the part after it for a clean output.
+    # Split the response at the first occurrence of "assistant" and return only the part after it
     response = response.split("assistant", 1)[1].strip()
     return f"\n{response}"
 
+# Example photos and prompts
+examples = [
+    {"image": "Cowboy Hat.jpg", "text": "Describe the photo"},
+    {"image": "Kynda Coffee.jpg", "text": "Search for the business name on his t-shirt to get a description of where the person is."},
+    {"image": "Norway.jpg", "text": "Where is this person?"}
+]
+
+# Load example images
+example_images = [Image.open(example["image"]) for example in examples]
+
 # Gradio
 interface = gr.Interface(
     fn=predict,
@@ -46,7 +56,8 @@ interface = gr.Interface(
     ],
     outputs=gr.Textbox(label="Output"),
     title="Llama 3.2 11B Vision Instruct Chat",
-    description="Image + text chat."
+    description="Image + text chat.",
+    examples=[{"image": image, "text": example["text"]} for image, example in zip(example_images, examples)]
 )
 
 interface.launch()