Alex Brooks 68a1d23e4d [Bugfix/Feature] Remove Hardcoded Flash Attention in Bagel & Support GQA in SDPA Backend (#3728) ...

Signed-off-by: Alex Brooks <albrooks@redhat.com>

2026-05-29 10:36:15 +08:00

..

ring

[Cleanup] Replace bare print() with logger and use specific exception types (#2228)

2026-04-03 23:59:07 +08:00

utils

[Feature] hunyuanimage support flash attn (#2981)

2026-05-10 22:08:24 +08:00

__init__.py

[Diffusion] Init Attention Backends and Selector for Diffusion (#115)

2025-12-05 10:46:20 +00:00

abstract.py

support online FP8 quantization for FA on NPU #2236 (#2640)

2026-05-13 11:46:00 +08:00

cudnn_attn.py

[Blackwell] Add CUDNN_ATTN and FLASHINFER_ATTN backends for diffusion (auto-route) (#3079)

2026-05-28 11:09:25 +08:00

flash_attn.py

[XPU] set flash_attn as default diffusion attn backend and fix k_len for cross_attn (#3525)

2026-05-18 23:58:15 +00:00

flashinfer_attn.py

[Blackwell] Add CUDNN_ATTN and FLASHINFER_ATTN backends for diffusion (auto-route) (#3079)

2026-05-28 11:09:25 +08:00

registry.py

[Blackwell][1/N] Add SageAttention3 diffusion backend on blackwell(GB200/B200/RTX5090/PRO6000/DGX Spark available) (#3015)

2026-05-28 14:08:45 +00:00

ring_flash_attn.py

[Bugfix] Raise ValueError when joint_strategy='rear' and causal=True in Ring Attention (#767)

2026-01-13 17:13:31 +08:00

ring_pytorch_attn.py

[CI] Refactor test_sequence_parallel.py and add a warmup run for more accurate performance stat (#1165)

2026-02-03 12:23:15 +08:00

sage_attn3.py

[Blackwell][1/N] Add SageAttention3 diffusion backend on blackwell(GB200/B200/RTX5090/PRO6000/DGX Spark available) (#3015)

2026-05-28 14:08:45 +00:00

sage_attn.py

[Diffusion][Attention] Support per-role attention backend via CLI (#2681)

2026-05-10 11:59:05 +08:00

sdpa.py

[Bugfix/Feature] Remove Hardcoded Flash Attention in Bagel & Support GQA in SDPA Backend (#3728)

2026-05-29 10:36:15 +08:00