Update main.py

main.py

@@ -145,7 +145,7 @@ def predict(image):
     # Clean up intermediates
     del model_inputs, generation
     clean_memory()
-    del model,processor
+    # del model,processor
     return decoded
 
 @app.post("/extract_text")
@@ -172,7 +172,7 @@ async def batch_extract_text(batch_size:int, background_tasks: BackgroundTasks,
         start_time = time.time()
         
         # Limit batch size for memory management
-        max_batch_size = batch_size  # Adjust based on your GPU memory
+        max_batch_size = 32  # Adjust based on your GPU memory
         
         # if len(files) > 32:
         #     return {"error": "A maximum of 20 images can be processed at a time."}