david/flash-attention: flash-attention from https://github.com/Dao-AILab/flash-attention @ fa3-kvcache-gqa

tükrözi: https://github.com/Dao-AILab/flash-attention

Tri Dao c7f32a8409 [CrossEntropy] Support precomputed LSE		3 hónapja
..
test_cross_entropy.py	c7f32a8409 [CrossEntropy] Support precomputed LSE	3 hónapja
test_cross_entropy_parallel.py	c7f32a8409 [CrossEntropy] Support precomputed LSE	3 hónapja