huggingface / nanotron

Minimalistic large language model 3D-parallelism training
Apache License 2.0
1.14k stars 107 forks source link

Bump flash-attn to 2.5.0 #40

Closed NouamaneTazi closed 8 months ago

NouamaneTazi commented 8 months ago

Make sure you pip install -U flash-attn