mlx>=0.29.2
numpy
transformers>=4.39.3
protobuf
pyyaml
jinja2

[evaluate]
lm-eval
tqdm

[test]
datasets
lm-eval

[train]
datasets
tqdm
