Update app.py

app.py

@@ -8,6 +8,7 @@ import os
 
 model_name = "meta-llama/Llama-3.2-11B-Vision-Instruct"
 token = os.getenv("HUGGINGFACE_TOKEN").strip()
+
 processor = AutoProcessor.from_pretrained(model_name, token=token)
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
@@ -27,15 +28,25 @@ def analyze_image(image, prompt):
             ]}
         ]
         input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
+        
         inputs = processor(
             image,
             input_text,
             add_special_tokens=False,
             return_tensors="pt"
         ).to(model.device)
+        
+        # Separate inputs for generate method
+        generate_inputs = {
+            k: v for k, v in inputs.items() 
+            if k not in ['pixel_values', 'aspect_ratio_ids', 'aspect_ratio_mask']
+        }
+        
         with torch.no_grad():
-            output = model.generate(**inputs, max_new_tokens=100)
+            output = model.generate(**generate_inputs, max_new_tokens=100)
+        
         result = processor.decode(output[0], skip_special_tokens=True)
+        
         try:
             return json.loads(result)
         except json.JSONDecodeError:
@@ -43,9 +54,7 @@ def analyze_image(image, prompt):
     except Exception as e:
         return {"error": str(e), "traceback": traceback.format_exc()}
 
-default_prompt = """Analyze this image and determine if it contains a data logger. 
-A data logger is typically a small, black electronic device used to monitor and record data 
-over time, such as voltage, temperature, or current, via external sensors.
+default_prompt = """Analyze this image and determine if it contains a data logger. A data logger is typically a small, black electronic device used to monitor and record data over time, such as voltage, temperature, or current, via external sensors.
 
 If a data logger is present in the image, respond with:
 {"present": true, "reason": "Brief explanation of why you believe it's a data logger"}
@@ -69,4 +78,77 @@ iface = gr.Interface(
     ]
 )
 
-iface.launch()
+iface.launch()
+
+# import torch
+# from PIL import Image
+# from transformers import AutoProcessor, AutoModelForCausalLM
+# import gradio as gr
+# import json
+# import traceback
+# import os
+
+# model_name = "meta-llama/Llama-3.2-11B-Vision-Instruct"
+# token = os.getenv("HUGGINGFACE_TOKEN").strip()
+# processor = AutoProcessor.from_pretrained(model_name, token=token)
+# model = AutoModelForCausalLM.from_pretrained(
+#     model_name,
+#     quantization_config={"load_in_4bit": True},
+#     token=token
+# )
+
+# if torch.cuda.is_available():
+#     model = model.to('cuda')
+
+# def analyze_image(image, prompt):
+#     try:
+#         messages = [
+#             {"role": "user", "content": [
+#                 {"type": "image"},
+#                 {"type": "text", "text": prompt}
+#             ]}
+#         ]
+#         input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
+#         inputs = processor(
+#             image,
+#             input_text,
+#             add_special_tokens=False,
+#             return_tensors="pt"
+#         ).to(model.device)
+#         with torch.no_grad():
+#             output = model.generate(**inputs, max_new_tokens=100)
+#         result = processor.decode(output[0], skip_special_tokens=True)
+#         try:
+#             return json.loads(result)
+#         except json.JSONDecodeError:
+#             return {"error": "Failed to parse model output as JSON", "raw_output": result}
+#     except Exception as e:
+#         return {"error": str(e), "traceback": traceback.format_exc()}
+
+# default_prompt = """Analyze this image and determine if it contains a data logger. 
+# A data logger is typically a small, black electronic device used to monitor and record data 
+# over time, such as voltage, temperature, or current, via external sensors.
+
+# If a data logger is present in the image, respond with:
+# {"present": true, "reason": "Brief explanation of why you believe it's a data logger"}
+
+# If no data logger is visible, respond with:
+# {"present": false, "reason": "Brief explanation of why you believe there's no data logger"}
+
+# Ensure your response is in valid JSON format."""
+
+# iface = gr.Interface(
+#     fn=analyze_image,
+#     inputs=[
+#         gr.Image(type="pil", label="Upload Image"),
+#         gr.Textbox(label="Prompt", value=default_prompt, lines=10)
+#     ],
+#     outputs=gr.JSON(label="Analysis Result"),
+#     title="Data Logger Detection using Llama 3.2 Vision",
+#     description="Upload an image and customize the prompt to check if it contains a data logger.",
+#     examples=[
+#         ["bad.png", default_prompt]
+#     ]
+# )
+
+# iface.launch()