download_models.py 946 B

123456789101112131415161718192021222324252627282930313233
  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/OCR/paddleocr_torch/*",
  10. # "models/TabRec/TableMaster/*",
  11. # "models/TabRec/StructEqTable/*",
  12. ]
  13. model_dir = snapshot_download(
  14. "opendatalab/PDF-Extract-Kit-1.0",
  15. allow_patterns=mineru_patterns,
  16. local_dir="/opt/",
  17. )
  18. layoutreader_pattern = [
  19. "*.json",
  20. "*.safetensors",
  21. ]
  22. layoutreader_model_dir = snapshot_download(
  23. "hantian/layoutreader",
  24. allow_patterns=layoutreader_pattern,
  25. local_dir="/opt/layoutreader/",
  26. )
  27. model_dir = model_dir + "/models"
  28. print(f"model_dir is: {model_dir}")
  29. print(f"layoutreader_model_dir is: {layoutreader_model_dir}")