david/flash-attention

réplica de https://github.com/Dao-AILab/flash-attention

Autor	SHA1 Mensaxe	Data
Tri Dao	71befc19e1 [Loss] Use flash_attn.losses.cross_entropy.CrossEntropyLoss	hai 1 ano
Tri Dao	dff68c2b22 Add smoothing for CrossEntropyParallel, rename to CrossEntropyLoss	%!s(int64=2) %!d(string=hai) anos
Tri Dao	0bf5e50038 Release training code	%!s(int64=2) %!d(string=hai) anos