smallq-flash-attention-ascend / smallq_flash_attention

Commit History

Replace with cleaned vector-only implementation
86d4a3b
verified

adasdadsd commited on

v2.0: Full float32 precision pipeline, headDim=64 support, 448/448 production tests passed
2145fce
verified

adasdadsd commited on

Delete smallq_flash_attention/op_kernel/smallq_flash_attention_impl.h with huggingface_hub
8737149
verified

adasdadsd commited on

Delete smallq_flash_attention/op_kernel/smallq_flash_attention.cpp with huggingface_hub
0615195
verified

adasdadsd commited on

Delete smallq_flash_attention/op_host/smallq_flash_attention_tiling.h with huggingface_hub
024861d
verified

adasdadsd commited on

Delete smallq_flash_attention/op_host/smallq_flash_attention_tiling.cpp with huggingface_hub
6f8d74e
verified

adasdadsd commited on

Delete smallq_flash_attention/op_host/smallq_flash_attention_proto.cpp with huggingface_hub
d41d2c1
verified

adasdadsd commited on

Delete smallq_flash_attention/op_host/smallq_flash_attention_def.cpp with huggingface_hub
87e9498
verified

adasdadsd commited on

Delete smallq_flash_attention/README.md with huggingface_hub
3400f18
verified

adasdadsd commited on

Delete smallq_flash_attention/CMakeLists.txt with huggingface_hub
ccf172b
verified

adasdadsd commited on

Upload folder using huggingface_hub
86e9bfe
verified

adasdadsd commited on