download_models.py 906 B

1234567891011121314151617181920212223242526272829303132
  1. #!/usr/bin/env python
  2. from huggingface_hub import snapshot_download
  3. if __name__ == "__main__":
  4. mineru_patterns = [
  5. # "models/Layout/LayoutLMv3/*",
  6. "models/Layout/YOLO/*",
  7. "models/MFD/YOLO/*",
  8. "models/MFR/unimernet_hf_small_2503/*",
  9. # "models/TabRec/TableMaster/*",
  10. # "models/TabRec/StructEqTable/*",
  11. ]
  12. model_dir = snapshot_download(
  13. "opendatalab/PDF-Extract-Kit-1.0",
  14. allow_patterns=mineru_patterns,
  15. local_dir="/opt/",
  16. )
  17. layoutreader_pattern = [
  18. "*.json",
  19. "*.safetensors",
  20. ]
  21. layoutreader_model_dir = snapshot_download(
  22. "hantian/layoutreader",
  23. allow_patterns=layoutreader_pattern,
  24. local_dir="/opt/layoutreader/",
  25. )
  26. model_dir = model_dir + "/models"
  27. print(f"model_dir is: {model_dir}")
  28. print(f"layoutreader_model_dir is: {layoutreader_model_dir}")