Update app.py

app.py

@@ -1,28 +1,43 @@
 import requests
 import torch
 from PIL import Image
-from transformers import MllamaForConditionalGeneration, AutoProcessor
+from transformers import LlamaForConditionalGeneration, AutoProcessor
 
+# Define the model ID, replace with the correct ID if needed
 model_id = "meta-llama/Llama-3.2-11B-Vision-Instruct"
 
-model = MllamaForConditionalGeneration.from_pretrained(
+# Load the model in bfloat16 or float16 if needed
+model = LlamaForConditionalGeneration.from_pretrained(
     model_id,
-    torch_dtype=torch.bfloat16,
-    device_map="auto",
+    torch_dtype=torch.bfloat16,  # Change to torch.float16 if hardware doesn't support bfloat16
+    device_map="auto",  # Automatically selects the appropriate device
 )
+
+# Load the processor
 processor = AutoProcessor.from_pretrained(model_id)
 
+# Define an image URL
 url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/0052a70beed5bf71b92610a43a52df6d286cd5f3/diffusers/rabbit.jpg"
+
+# Fetch the image using requests
 image = Image.open(requests.get(url, stream=True).raw)
 
+# Define the messages in a format the model understands (adjust as needed)
 messages = [
     {"role": "user", "content": [
-        {"type": "image"},
-        {"type": "text", "text": "If I had to write a haiku for this one, it would be: "}
+        {"type": "image"},  # This indicates that the input contains an image
+        {"type": "text", "text": "Can you please describe this image in one sentence?"}
     ]}
 ]
+
+# Generate input text with the processor
 input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
+
+# Process the image and input text, prepare them for the model
 inputs = processor(image, input_text, return_tensors="pt").to(model.device)
 
-output = model.generate(**inputs, max_new_tokens=30)
-print(processor.decode(output[0]))
+# Run the model to generate a response
+output = model.generate(**inputs, max_new_tokens=70)
+
+# Decode and print the output
+print(processor.decode(output[0][inputs["input_ids"].shape[-1]:]))