pyyaml>=6.0
requests>=2.28

[gguf]
llama-cpp-python>=0.2.70

[hf]
transformers>=4.40
torch>=2.2
datasets>=2.14
scikit-learn>=1.2
accelerate>=0.30
