Commit Graph

9209 Commits

Author SHA1 Message Date
Alison Shao
7b22b8ff8a Fix sgl-kernel install: fail instead of PyPI fallback when artifacts missing (#17728) 2026-01-26 11:46:49 +08:00
chenxu214
444b9521e4 [Bugfix]Repeated add modelslim quant_config and bugfix with "enable-piecewise-cuda-graph" on NPU (#17511) 2026-01-26 09:51:07 +08:00
Simo Lin
46bc53a81b [smg] import db crate to replace self managed one (#17727) 2026-01-25 16:54:52 -08:00
Simo Lin
52d0ca944b [smg] import official wfaas crate to replace self managed one (#17724) 2026-01-25 15:20:58 -08:00
Kangyan-Zhou
592603d77b Fix flaky streaming logprobs test by handling detokenizer text buffering (#17687)
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
2026-01-25 15:09:06 -08:00
Simo Lin
bf139af37a [smg] use already published auth crate for better compilation speed (#17723) 2026-01-25 14:52:42 -08:00
Kangyan-Zhou
344eeaee90 Upload nightly test metrics to GH artifacts (#17696) 2026-01-25 14:35:14 -08:00
Simo Lin
97a36a72b7 [smg] remove dead tokenizer code (#17722) 2026-01-25 14:04:48 -08:00
Praneth Paruchuri
fc7096f80b [model-gateway] Optimize special token search using Aho-Corasick (#17387)
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2026-01-25 13:54:15 -08:00
Simo Lin
f5ac1ca10b [smg] use official tokenizer crate instead of manually built one (#17721) 2026-01-25 13:35:16 -08:00
Simo Lin
2c1e2674cc [misc] replace existing tool call code with new crate package (#17720) 2026-01-25 13:07:39 -08:00
Simo Lin
48f5c46a8d [misc] remove tool parser and tree benchmark as they are not meaningful atm (#17719) 2026-01-25 13:03:06 -08:00
Kangyan-Zhou
7ca8c12e0e Extend b200 kernel tests timeout for CPU differences (#17718) 2026-01-25 12:59:39 -08:00
Simo Lin
86c7bc6431 [smg] update crate for tools (#17717) 2026-01-25 12:56:42 -08:00
Kangyan-Zhou
8d3e1ac0c8 Add an all type in pyproject.tml to include diffusion support (#17697) 2026-01-25 12:52:13 -08:00
Kangyan-Zhou
9123491430 A few updates to the night tests (#17694) 2026-01-25 11:20:17 -08:00
HandH1998
a883906a24 Support mxint4 flashinfer_trtllm moe gemm (#16892) 2026-01-26 00:15:53 +08:00
Mick
b105dad5da [diffusion] refactor: remove useless lazy-import cache-dit codes (#17659) 2026-01-25 22:43:22 +08:00
Simo Lin
7a1d7ab47b remove self managed protocols as it has been replaced with official oai spec (#17711) 2026-01-25 06:38:13 -08:00
Simo Lin
8db2802b2d update to use official openai protocol crate (#17710) 2026-01-25 06:35:59 -08:00
Simo Lin
6af22f8dbf use published reasoning parser crate (#17709) 2026-01-25 06:15:03 -08:00
Zhengbo Wang
fb61164f27 [Refactor] Use is_in_ci() utility in JIT kernel benchmarks (#17118) 2026-01-25 20:40:47 +08:00
Makcum888e
64d809937a revert row from https://github.com/sgl-project/sglang/pull/17584/ (#17701) 2026-01-25 12:17:47 +03:00
Makcum888e
d1042e0d62 [Refactore] [CI] Remove redundant CI test runs step 2 (#17584) 2026-01-24 23:39:48 -08:00
xjx471258437
9bd92ba0f6 Support PD disaggregation with different TP/DP size for Qwen3-Next (#16056)
Co-authored-by: xjx392321 <xjx392321@alibaba-inc.com>
2026-01-25 15:34:02 +08:00
Ke Bao
30ece5e1d6 Fix swa memory pool size with spec (#17630) 2026-01-25 14:10:43 +08:00
Mohammad Miadh Angkad
1674b9ef44 [DeepSeek-V3.2] Fix TRT-LLM NSA in target_verify/draft_extend (#17662) 2026-01-25 13:10:14 +08:00
Alison Shao
9121f22656 Add PyTorch .bin file validation to CI weight validation (#17533) 2026-01-24 19:18:15 -08:00
Kangyan-Zhou
b829b797ef Fix slash command handler trigger condition by trimming the comments (#17691) 2026-01-24 18:54:46 -08:00
Chen Shen
59f027a8c8 [diffusion]: Fix ZImage SP sharding for caption and latent (#17301)
Co-authored-by: rhyshen <rhyshen@tencent.com>
Co-authored-by: florianzhao <florianzhao@tencent.com>
2026-01-25 10:10:48 +08:00
Xinyuan Tong
37c04c2245 fix: Refactor register_image_processor to use kwarg instead of positional arg (#17685)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2026-01-24 15:31:01 -08:00
Kangyan-Zhou
69a7a70e47 Temporarily disable lora overlap loading test due to flakiness (#17683) 2026-01-24 12:24:44 -08:00
Kangyan-Zhou
137eb5b95c Fix NSA indexer test and move it to pre commit test (#17682) 2026-01-24 12:06:18 -08:00
Kangyan-Zhou
8656a146a6 Fix test timeout issue in pr-test (#17681) 2026-01-24 11:46:08 -08:00
Trevor Morris
2c2c4e446b [NVIDIA] Add flashinfer all-to-all MOE dispatcher (#14668) 2026-01-24 22:59:55 +08:00
TMC
458a43d4ac [NPU] torch_npu profiler tensorboard path type fix (#17545) 2026-01-24 22:55:49 +08:00
Yuan Luo
0c8165ffbd [Kimi-Linear] Refactor Kimi-Linear to support RadixLinearAttention (#17506)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2026-01-24 21:27:13 +08:00
yunkchen
bf19d20d89 [Bugfix] fix TypeError when log-requests-level >=2 in prefill node warmup (#17129) 2026-01-24 19:16:22 +08:00
Lianmin Zheng
0834f9afeb [Auto Sync] Update test_deterministic.py (20260124) (#17665)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Jiayi Yuan <34369239+jy-yuan@users.noreply.github.com>
2026-01-24 02:52:47 -08:00
Glen Liu
a6280b2a23 add documentation example for LoRA overlap loading and cleanup unused function (#17464) 2026-01-24 15:33:16 +08:00
Xiaoyu Zhang
3992a023e6 Move fa4 from sgl-kernel to jit kernel (#17353) 2026-01-24 15:25:03 +08:00
Xiaoyu Zhang
7a4bb0d516 [Diffusion] Add diffusion time embedding to jit kernel (#17658) 2026-01-24 14:27:08 +08:00
Ke Bao
fb683be6eb Use attn tp group in embedding for more models (#17570) 2026-01-24 13:37:44 +08:00
strgrb
176da1bbdd Fix: mistake sigmoid in kda (#17508) 2026-01-24 13:35:14 +08:00
Qi Yuhang
4c512a7d1d [JIT Kernel]Add Some CUDA Runtime API Wrapper for JIT Kernel Header (#17588) 2026-01-24 12:57:58 +08:00
Douglas Yang
58799d91a7 fix: nightly wheel naming for non-post versions (#17538) 2026-01-24 12:51:52 +08:00
GMI Xiao Jin
d0919be733 [diffusion] model: LTX-2 Support (2/2) (#17496)
Co-authored-by: Fan Yin <1106310035@qq.com>
Co-authored-by: Yuhao Yang <47235274+yhyang201@users.noreply.github.com>
2026-01-24 12:51:37 +08:00
GMI Xiao Jin
797a9811a2 [diffusion] model: LTX-2 (1/2) (#17495)
Co-authored-by: FlamingoPg <1106310035@qq.com>
2026-01-24 11:59:48 +08:00
Ananya
894928a951 Refactor: Extract DeepSeek common utilities into shared module (#16969) 2026-01-24 11:29:52 +08:00
Baizhou Zhang
0dfe46dafb [Docker] Install cudnn==9.16 for cuda 13 image to avoid check error (#17668) 2026-01-24 11:27:03 +08:00