ops.aten._scaled_dot_product_flash_attention_for_cpu(query, key, value, dropout_p, is_causal, attn_mask=atten_mask, scale=scale) Output: Floating point exception (core dumped) ASAN report: AddressSanitizer:DEADLYSIGNAL === ==3698429==ERROR: AddressSanitizer: FPE on unknown address 0x7f64779...
This branch is 86 commits behind Dao-AILab/flash-attention:main.Folders and files Latest commit XiaobingSuper don't save inputs buffer of FlashAttenFunc to reduce memory usage for… 0dfb281· Dec 12, 2024 History733 Commits .github/workflows [CI] Use MAX_JOBS=1 with nvcc 12.3, don't ...
Makefile allow for uploading to pypi Nov 16, 2022 README.md Add how to import FA3 to documentation (Dao-AILab#1112) Dec 6, 2024 setup.py [CI] Use MAX_JOBS=1 with nvcc 12.3, don't need OLD_GENERATOR_PATH Dec 8, 2024 usage.md Update usage.md Jul 15, 2023 Repository files naviga...
This branch is 90 commits behind Dao-AILab/flash-attention:main.Folders and files Latest commit XiaobingSuper don't save inputs buffer of FlashAttenFunc to reduce memory usage for… 0dfb281· Dec 12, 2024 History733 Commits .github/workflows [CI] Use MAX_JOBS=1 with nvcc 12.3, don't ...
This branch is 86 commits behind Dao-AILab/flash-attention:main.Folders and files Latest commit XiaobingSuper don't save inputs buffer of FlashAttenFunc to reduce memory usage for… 0dfb281· Dec 12, 2024 History733 Commits .github/workflows [CI] Use MAX_JOBS=1 with nvcc 12.3, don't ...
This branch is 86 commits behind Dao-AILab/flash-attention:main.Folders and files Latest commit XiaobingSuper don't save inputs buffer of FlashAttenFunc to reduce memory usage for… 0dfb281· Dec 12, 2024 History733 Commits .github/workflows [CI] Use MAX_JOBS=1 with nvcc 12.3, don't ...