einops
hf_transfer
huggingface_hub
mlx
numpy
torch
transformers>=4.35