Training a 1.93B byte-level LM on one A100: 60K steps, flat to 4x context

 (huggingface.co)

1 points | by EverestAn 5 hours ago ago

1 comments