Sep 3

1. 21:28

   Streamline prompt for fast sub-5s LLM evaluation

   pareshmishra
   fab4da3
2. 21:27

   Optimize LLM latency by disabling chain-of-thought generation for direct JSON extraction

   pareshmishra
   bcb01d9
3. 20:54

   Update default model to nvidia/nemotron-3-nano-omni-30b-a3b-reasoning

   pareshmishra
   d70b0a7
4. 20:48

   Gracefully display OCR extracted text when LLM encounters authentication errors

   pareshmishra
   7b48434
5. 20:42

   Add headless OpenCV and detailed OCR error tracking

   pareshmishra
   0de8787
6. 20:41

   Optimize OCR pre-warming and RapidOCR execution order

   pareshmishra
   24d02e7
7. 20:39

   Support numpy and dict image formats from Gradio in OCR service

   pareshmishra
   3f1abd8
8. 20:30

   Fix os import and PaddleOCR initialization compatibility

   pareshmishra
   a18b90b
9. 20:05

   Build OCR document intelligence V1

   pareshmishra
   7c03a86
10. 19:54

    Initial AI Document Intelligence OCR implementation

    pareshmishra
    8e798a2