flash-attention/tests
Tri Dao 9d3116addf Don't enforce bitwise consistency for dq in race condition test
Since we could be parallelizing over seqlen_k
2022-11-13 12:21:51 -08:00
..
losses Add fused cross entropy loss 2022-11-12 21:58:41 -08:00
test_flash_attn.py Don't enforce bitwise consistency for dq in race condition test 2022-11-13 12:21:51 -08:00
test_rotary.py Implement rotary embedding in CUDA 2022-11-04 22:42:01 -07:00