Yuan Luoandluoyuan.luo d1ec93e3ac Optimize layernorm_gated for Qwen3-Next (#16397) ...

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>

2026-01-10 20:55:31 +08:00

..

fla

Optimize layernorm_gated for Qwen3-Next (#16397)

2026-01-10 20:55:31 +08:00

mamba

[nemetron/mtp] fix nemotron mtp (#16275)

2026-01-09 23:22:21 +08:00

nsa

[DSv32] Overlap indexer weights_proj during dual_stream decode (#16637)

2026-01-10 13:06:44 +08:00

triton_ops

[refactor] Move trtllm_fp8_kv_kernel to triton_ops directory (#15044)

2025-12-14 15:22:37 -08:00

wave_ops

…

aiter_backend.py

[AMD] Fix aiter page-size handling, DeepSeek MLA tuple inputs, and HiCache/FA3 decode-backend override (#16531)

2026-01-07 21:14:32 -08:00

attention_registry.py

[NPU][1/N] NPU basic functions refactor and new modelslim quant type (#13359)

2025-12-04 16:15:31 +08:00

base_attn_backend.py

[Ascend] Support enable-mixed-chunk in non-MLA scenarios (#12491)

2025-11-27 00:00:21 +08:00

cutlass_mla_backend.py

…

double_sparsity_backend.py

…

dual_chunk_flashattention_backend.py

…

flashattention_backend.py

Only allocate encoder metadata for encoder-decoder models (#16527)

2026-01-06 22:17:04 -08:00

flashinfer_backend.py

Move swa memory pool to a seperate file (#16347)

2026-01-04 22:39:30 +08:00

flashinfer_mla_backend.py

Piecewise Cuda Graph Support for gpt-oss model (#13045)

2025-11-15 09:28:00 +08:00

flashmla_backend.py

Fix: fix flashmla fp8 kv cache acc error (#13841)

2025-11-30 13:38:19 -08:00

hybrid_attn_backend.py

…

hybrid_linear_attn_backend.py

support page size large than 64 for mamba radix cache with fix (#16768)

2026-01-09 13:28:37 +08:00

intel_amx_backend.py

…

merge_state.py

…

nsa_backend.py

Clean Some Environment Variables for DeepSeek V32 (#15938)

2026-01-07 14:00:16 +08:00

tbo_backend.py

fix ds3.2 nsa backend prefill TBO (#14901)

2025-12-21 13:16:46 -08:00

torch_flex_backend.py

…

torch_native_backend.py

Support kv8 (FP8) with torch_native attention backend (#12596)

2025-12-28 10:48:30 +08:00

triton_backend.py

feat(SpecEagleV2): add standalone_worker_v2 (#12625)

2025-12-30 17:55:04 +08:00

trtllm_mha_backend.py

Feature/trtllm mha workspace size configurable #15089 (#15131)

2025-12-17 17:07:23 -08:00

trtllm_mla_backend.py

[MTP][spec_v2] Fix TRTLLM MLA backend crash in EAGLE draft_extend mode (#15790)

2026-01-10 19:58:23 +08:00

utils.py

Fix gpu-fault when running mtp in eager mode (#15233)

2025-12-17 01:25:06 -08:00

vision_utils.py

…

vision.py

[jit kernel] support dtype as a cpp template parameter (#16452)

2026-01-08 13:54:33 +08:00

wave_backend.py

…

xpu_backend.py

[Auto Sync] Rename is_hybrid to is_hybrid_swa (#14252)

2025-12-01 23:24:24 -08:00