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