Ying Zhang 496fdc4f6c Add seqused_q in fwd / bwd and seqused_k in bwd. 4 miesięcy temu
..
layers 3566596ad8 Fix typo in RotaryEmbedding forward output type (#666) 1 rok temu
losses ec6d22143b [CrossEntropy] Change ignored_index -> ignore_index 9 miesięcy temu
models 0d810cfb73 Fix KeyError handling for non-existing key in state_dict.pop() (#898) 6 miesięcy temu
modules 3f1b4d38e7 Fix: check the type of max_seqlen_k instead of checking max_seqlen twice (#1127) 5 miesięcy temu
ops d79f9b41a8 [CrossEntropy] Use online softmax to simplify implementation 5 miesięcy temu
utils 320fb59487 Update citation 8 miesięcy temu
__init__.py 418d677192 Bump to v2.6.3 6 miesięcy temu
bert_padding.py 496fdc4f6c Add seqused_q in fwd / bwd and seqused_k in bwd. 4 miesięcy temu
flash_attn_interface.py ef3e358a25 remove lambda (#1056) 6 miesięcy temu
flash_attn_triton.py f1a73d0740 Run isort and black on python files 1 rok temu
flash_attn_triton_og.py f1a73d0740 Run isort and black on python files 1 rok temu
flash_blocksparse_attention.py f1a73d0740 Run isort and black on python files 1 rok temu
flash_blocksparse_attn_interface.py f1a73d0740 Run isort and black on python files 1 rok temu
fused_softmax.py f1a73d0740 Run isort and black on python files 1 rok temu
pyproject.toml 73bd3f3bbb Move pyproject.toml to flash-attn and tests dir to avoid PEP 517 1 rok temu