david/flash-attention: flash-attention from https://github.com/Dao-AILab/flash-attention

SueJane 3f1b4d38e7 Fix: check the type of max_seqlen_k instead of checking max_seqlen twice (#1127)		4 сар өмнө
..
__init__.py	ece539abd6 Add __init__.py files to subdirectories for installation	2 жил өмнө
block.py	abbc131173 [LayerNorm] Switch from CUDA to Triton implementation	11 сар өмнө
embedding.py	f1a73d0740 Run isort and black on python files	1 жил өмнө
mha.py	3f1b4d38e7 Fix: check the type of max_seqlen_k instead of checking max_seqlen twice (#1127)	4 сар өмнө
mlp.py	c3b2196652 Add Alibi to MHA, test with Baichuan-13B	11 сар өмнө