Hugging Face's logo

Datasets:
uv-scripts
/
ocr

Tags:
uv-script
ocr
extraction
vision-language-model
document-processing
hf-jobs
Dataset card Files
xet
 Community
12
ocr

Commit History

  1. Fix: Constrain pyarrow to <18.0.0 for datasets compatibility

    davanstrien HF Staff
    9e458ca

  2. Fix: Add pyarrow dependency for datasets compatibility

    davanstrien HF Staff
    57aa59f

  3. Add DeepSeek-OCR-2 script with Visual Causal Flow architecture (testing)

    davanstrien HF Staff
    ee2a300

  1. Update PaddleOCR-VL configuration for vLLM compatibility and enhance model parameters

    davanstrien HF Staff
    c7d345d

  2. Add pyarrow to dependencies in paddleocr-vl.py

    davanstrien HF Staff
    9be50d6

  3. Refactor dependencies and clean up whitespace in paddleocr-vl.py

    davanstrien HF Staff
    dbf8967

  4. Add PaddleOCR-VL script for document processing with vLLM support

    davanstrien HF Staff
    76273a4

  1. Match official demo message format: text before image⁠

    davanstrien +1: Claude HF Staff
    ddf6b7f

  2. Update LightOnOCR target resolution to 1540px⁠

    davanstrien +1: Claude HF Staff
    6042172

  1. Add LightOnOCR documentation to README⁠

    davanstrien +1: Claude HF Staff
    a77e763

  2. Fix triton-kernels dependency for vLLM nightly⁠

    davanstrien +1: Claude HF Staff
    d805bea

  1. Update LightOnOCR script to use vLLM nightly wheels⁠

    davanstrien +1: Claude HF Staff
    94823db

  2. Add LightOnOCR script for fast document OCR⁠

    davanstrien +1: Claude HF Staff
    0e62664

  3. keep viewer

    davanstrien HF Staff
    916ff8e

  4. Critical quality fixes: temperature 0.1 and image resizing to 1288px⁠

    davanstrien +1: Claude HF Staff
    ea32fa8

  5. Fix message content order: text before image⁠

    davanstrien +1: Claude HF Staff
    a8ad6e1

  6. Add repetition penalty, guided decoding, and finish_reason logging⁠

    davanstrien +1: Claude HF Staff
    9e98db3

  7. Namespace metadata column by output column name⁠

    davanstrien +1: Claude HF Staff
    ee028d2

  8. Add --output-column parameter to olmocr2-vllm.py⁠

    davanstrien +1: Claude HF Staff
    ed67b34

  9. Change olmOCR2 temperature from 0.1 to 0.0 to match original implementation⁠

    davanstrien +1: Claude HF Staff
    c588c51

  10. Add olmOCR2 script and usage instructions

    davanstrien HF Staff
    8388345

  1. Remove embedded DeepSeek-OCR git repository

    davanstrien HF Staff
    70d9d22

  2. Document DeepSeek-OCR in README with real examples⁠

    davanstrien +1: Claude HF Staff
    2d80026

  3. Add prompt mode presets and remove test script⁠

    davanstrien +1: Claude HF Staff
    3073a00

  4. prompt

    davanstrien HF Staff
    88db448

  5. Add UV_TORCH_BACKEND=auto to HF Jobs example⁠

    davanstrien +1: Claude HF Staff
    cf61cef

  6. Fix vLLM nightly wheel index URL⁠

    davanstrien +1: Claude HF Staff
    d2ade5c

  7. Simplify HF Jobs token syntax to -s HF_TOKEN⁠

    davanstrien +1: Claude HF Staff
    2db2034

  8. Switch DeepSeek-OCR scripts to vLLM nightly wheels⁠

    davanstrien +1: Claude HF Staff
    90889e4

  9. Add DeepSeek-OCR vLLM scripts (PR #27247 testing)

    davanstrien HF Staff
    f987b04

  1. Fix output_path parameter in model.infer⁠

    davanstrien +1: Claude HF Staff
    e69f406

  2. Use local temp_images directory instead of system temp⁠

    davanstrien +1: Claude HF Staff
    a9bc1eb

  3. Fix temp file handling in DeepSeek-OCR⁠

    davanstrien +1: Claude HF Staff
    123f376

  4. Pin transformers version for DeepSeek-OCR compatibility⁠

    davanstrien +1: Claude HF Staff
    dace8e2

  5. Fix DeepSeek-OCR dependencies⁠

    davanstrien +1: Claude HF Staff
    b2a28f3

  6. Add DeepSeek-OCR script (Transformers-based)⁠

    davanstrien +1: Claude HF Staff
    fab5e6e

  1. Add Nanonets-OCR2 documentation to README⁠

    davanstrien HF Staff
    7593b9a

  2. Simplify nanonets-ocr2.py to 3B-only and fix dataset card⁠

    davanstrien HF Staff
    fa7a872

  3. Add Nanonets-OCR2 script with dual model support (1.5B/3B)⁠

    davanstrien HF Staff
    c0663cf

  1. Add DoTS.ocr to documentation with official --secrets syntax

    davanstrien HF Staff
    90ace90

  2. Fix dataset card: remove viewer:false, remove speed claim, simplify features

    davanstrien HF Staff
    2204718

  3. Update dots-ocr.py with official prompts and simplified design

    davanstrien HF Staff
    35c7121

  1. Add NuMarkdown to README with HF Jobs example

    davanstrien HF Staff
    7165fc4

  2. Fix dataset card push with whoami for repo ID resolution

    davanstrien HF Staff
    a7e4157

  3. Add NuMarkdown-8B-Thinking OCR script with reasoning capabilities

    davanstrien HF Staff
    b366864

  1. Remove dots-ocr from README due to compatibility issues⁠

    davanstrien HF Staff
    3f59035

  2. Add RolmOCR script and enhance OCR scripts with new features⁠

    davanstrien HF Staff
    43324c8

  1. Optimize default settings based on performance testing⁠

    davanstrien HF Staff
    cea7723

  2. Update README.md

    davanstrien HF Staff
    Verified signature
    b3cfa7b

  1. Update README.md to clarify that the script processes the first 10 images from the dataset

    davanstrien HF Staff
    3ef9d1f