Update llm.py

llm.py

@@ -36,9 +36,9 @@ processor=AutoProcessor.from_pretrained("datalab-to/chandra-ocr-2")
 model=AutoModelForImageTextToText.from_pretrained("datalab-to/chandra-ocr-2",quantization_config=quant_config)
 
 def execute_llm(model, processor,image, prompt: str):
-    
+    print('execulte llm')
     # Prompt (IMPORTANT)
-    prompt = "Extract all text from the image and return structured JSON with text and bounding boxes."
+    prompt = "Extract all text from the image and return structured JSON "
     
     messages = [
         {
@@ -58,13 +58,13 @@ def execute_llm(model, processor,image, prompt: str):
         return_dict=True,
         return_tensors="pt"
     ).to(model.device)
-    
+    print('input generated')
     # Generate
     outputs = model.generate(
         **inputs,
         max_new_tokens=1000
     )
-    
+    print('output bhi')
     # Decode only new tokens
     result = processor.decode(
         outputs[0][inputs["input_ids"].shape[-1]:],
@@ -72,11 +72,13 @@ def execute_llm(model, processor,image, prompt: str):
     )
     
     print(result)
+    return result
 # ✅ SYNC function (important)
 
 
 # ✅ ASYNC wrapper
 async def call_llm(image,prompt: str):
+    print('call llm')
     loop = asyncio.get_event_loop()
 
     func = partial(execute_llm, model, processor,image, prompt)