Jeff Huang e1af9b7afb 1. Relax kv_blockscale page_size restriction from == 1024 to >= kN0 ...

2. Rename QScaleKargsSelector -> GetQScaleKargs for naming consistency
3. Remove unused BlockAttentionQuantScaleEnumToStr<KV_BLOCKSCALE>

2026-02-03 11:18:14 +08:00

..

__init__.py

chore(copyright): update copyright header for example directory (#3273)

2025-11-24 18:02:41 -08:00

fmha_batch_prefill.py

1. Relax kv_blockscale page_size restriction from == 1024 to >= kN0

2026-02-03 11:18:14 +08:00

fmha_bwd.py

chore(copyright): update copyright header for example directory (#3273)

2025-11-24 18:02:41 -08:00

fmha_fwd_appendkv.py

chore(copyright): update copyright header for example directory (#3273)

2025-11-24 18:02:41 -08:00

fmha_fwd_splitkv.py

Add attention sink support for FMHA FWD (#3368)

2025-12-15 12:21:59 +08:00

fmha_fwd.py

Revert "Revert " Fp8 block scale quantization for fmha fwd (#3330)" (#3633)" (#3635)

2026-01-23 09:03:22 -08:00

fmha_pagedkv_prefill.py

Add attention sink support for FMHA FWD (#3368)

2025-12-15 12:21:59 +08:00