Commit Graph

9353 Commits

Author SHA1 Message Date
jeff
04efd03dbf Fix OOM in DeepSeek weight loading by deferring dict(weights) materialization (#17744) 2026-01-31 13:59:00 +08:00
Yifan Cui
45fe51a28e Reduce topk kernel shared memory from 128KB to 32KB for better occupancy (#17747)
Co-authored-by: Claude <noreply@anthropic.com>
2026-01-30 21:42:21 -08:00
Hudson Xing
c72bf50706 add reasoning_tokens usage test for tool call (#18022) 2026-01-30 21:09:23 -08:00
husf
c52578c7fd 【docs】【NPU】Update Expert Parallelism docs for Ascend NPU (#17940) 2026-01-30 23:42:11 -05:00
R0CKSTAR
dc77defdd0 Fix .gitignore may ignore files like core_attention.py (#18021)
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
2026-01-30 20:33:32 -08:00
Mohammad Miadh Angkad
d0d9cecd1b Fix cuBLAS >=12.9 detection for cu12/cu13 package naming (#17766) 2026-01-31 12:01:52 +08:00
Xiaoyu Zhang
22aad4e2c4 [Diffusion] Fix FLUX.1-schnell time embedding argument mismatch (#17988) 2026-01-31 11:47:27 +08:00
kk
ec76c390c9 Add ROCm + Mori docker build instructions in rocm.Dockerfile (#18018)
Co-authored-by: wunhuang <wunhuang@amd.com>
2026-01-30 19:12:07 -08:00
22dimensions
ee3058c6e8 [NPU] fix sgl-kernel-npu package url error in npu.Dockerfile (#18017)
Signed-off-by: 22dimensions <waitingwind@foxmail.com>
2026-01-31 11:06:58 +08:00
Tiance Wang
f6a4ff718f doc update for CANN version (#18014)
Co-authored-by: wangtiance <tiancew@qq.com>
2026-01-30 21:10:23 -05:00
Bi Xue
5d00150e99 [sglang] fix mm token padded value overlap with text token id (#17781) 2026-01-30 17:09:13 -08:00
JiaruiChang5268
e86476acfc [NPU] support llama-3.2-11B-vision-instruct mode for NPU (#17492)
Co-authored-by: McZyWu <zhuoyun.wu.23@ucl.ac.uk>
Co-authored-by: chenyang08056032 <chenyang08056032@163.com>
Co-authored-by: Hexq0210 <893781835@qq.com>
2026-01-31 08:49:38 +08:00
Siyuan Chen
578b119bc6 [BugFix] Fix server crashes when req.grammar and ngram spec are enabled (#17585) 2026-01-30 11:57:42 -08:00
Sam (Kesen Li)
81449b4bee Optimize GDN decode for Qwen3 Next (#17094) 2026-01-31 01:02:12 +08:00
Xiaoyu Zhang
abf13ccc11 [Diffusion] Fix lora default lora_scale bug (#17982) 2026-01-30 22:04:54 +08:00
Zheng Li
0c5a81acb8 [BUGFIX] Fix dp size > 1 for qwen3 vl model (#17624)
Co-authored-by: yizhang2077 <1109276519@qq.com>
2026-01-30 20:44:25 +08:00
Changhun Lee
c04efe030a [Model] Add K-EXAONE model support (#16294)
Signed-off-by: lkm2835 <lkm2835@gmail.com>
Co-authored-by: lgai-exaone <exaonemodels@lgresearch.ai>
Co-authored-by: lkm2835 <lkm2835@gmail.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2026-01-30 20:01:14 +08:00
amote-i
9c2a468e2c update ascend docs (#17987) 2026-01-30 18:03:55 +08:00
Fan Yin
8ce9609fa2 fix: fix SHM pointer re-serialization in DP attention (#17930) 2026-01-30 17:03:30 +08:00
Ke Bao
77a27e728c Add cuda graph status to prefill log (#17836) 2026-01-30 16:56:53 +08:00
Haotong Zhang
c8dc543dc5 SGLang Tracing: Improve root span attributes (#17008)
Signed-off-by: zhanghaotong <zhanghaotong.zht@antgroup.com>
2026-01-30 16:02:05 +08:00
Kangyan-Zhou
a16aba8340 Increase install dependency for gb200 (#17977) 2026-01-29 23:55:25 -08:00
jianzhao-xu
96584ab692 adapt MODELSCOPE download (#17922) 2026-01-29 23:26:54 -08:00
McZyWu
70db3398d1 [NPU] enhance accuracy for model kimi-vl-a3b-instruct (#17480)
Co-authored-by: cy <chenyang08056032@163.com>
2026-01-30 15:19:42 +08:00
cswuyg
33c053c50c fix(benchmark): add missing args for speculative decoding benchmark (#17974) 2026-01-29 23:05:42 -08:00
jianan-gu
c35aa0238c [CPU][INT4] Add INT4 kernels for CPU (#8226)
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2026-01-29 22:30:13 -08:00
Ma Mingfei
88f7759402 [CPU] optimize flash_attn_varlen_func (#15708) 2026-01-29 22:07:05 -08:00
jianan-gu
336dc4579e [CPU] Optimize Qwen3-next model on CPU (#12525)
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>
Co-authored-by: Fan Yin <1106310035@qq.com>
2026-01-29 22:03:58 -08:00
Polisetty V R K Jyothendra Varma
71e4d3b6bc [Intel GPU] fix import error to run DeepSeek-V2-Lite model with BF16 on XPU (#10858) 2026-01-29 21:53:53 -08:00
gaopengff
7541da15d2 Fix prefill latency performance drop of bench serving (#14592) 2026-01-29 21:28:17 -08:00
Polisetty V R K Jyothendra Varma
858dc80aff [Intel GPU] fix device in DeepseekScalingRotaryEmbedding to run DeepSeek-V2-Lite BF16 on XPU (#10021)
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>
2026-01-29 21:21:38 -08:00
LHXuuu
0e4d9ddbd6 Fix the scenario where eh_proj is quantized in the bailing moe nextn weights (#17808)
Signed-off-by: LHXuuu <xulianhao.xlh@antgroup.com>
2026-01-29 21:11:47 -08:00
Kangyan-Zhou
606ff09ef8 [Fix] Remove unused Type import in gpt_j.py (#17975)
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-29 21:11:11 -08:00
Koushik Dutta
632c7afa8c [Fix] add block size logic for sm120 smem size (#14311) 2026-01-29 21:01:57 -08:00
Wenchen Lo
046b29be16 GPTJForCausalLM Support (#7839)
Co-authored-by: b8zhong <b8zhong@uwaterloo.ca>
2026-01-29 21:00:04 -08:00
b8zhong
22df62d586 add weightless qk norm to RMSNorm interface for Llama 4 (#12813)
Co-authored-by: Brayden Zhong <b8zhong@users.noreply.github.com>
2026-01-29 19:09:55 -08:00
baonudesifeizhai
84ab611af8 model: support DeepSeek-OCR-2 (#17897) 2026-01-30 09:49:51 +08:00
Kangyan-Zhou
2cd2c3118d Add concurrency tracking to runner utilization report (#17963)
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-29 17:31:55 -08:00
YC Tseng
b7b1ba329e [AMD] fix pip sglang version (#17950) 2026-01-29 17:19:57 -08:00
StonyPort
2b3408ff14 feat: add forward timeout (#17831)
Co-authored-by: qiuxuan.lzw <qiuxuan.lzw@alibaba-inc.com>
2026-01-30 08:52:29 +08:00
Cheng Wan
6a6b36367e Fix logprob_start_len handling for prefill-only requests (#17395) 2026-01-29 15:14:43 -08:00
Kangyan-Zhou
c3bf53c7c1 Fix ci weight validation logic to check the safetensor completeness (#17917) 2026-01-29 13:00:42 -08:00
Alison Shao
1f75c2af4d Fix /tag-and-rerun-ci to do full rerun when PR has sgl-kernel changes (#17729) 2026-01-29 12:54:30 -08:00
Cheng Wan
a416af4be7 Fix capture_sizes range for pcg (#17956) 2026-01-29 12:46:35 -08:00
EduardDurech
1b6798a6a4 Fix torch.__version__ for PEP440 (#15682) 2026-01-29 11:55:13 -08:00
R0CKSTAR
4f2b73baf6 [MUSA] Add labeler config (#17923)
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
2026-01-29 09:58:54 -08:00
Hudson Xing
d417c6809e Add tool call tests for DeepSeek V3.2 in nightly CI (#17951) 2026-01-29 09:50:54 -08:00
Ratish P
88fb927cc9 [diffusion]: add dummy device attribute to fix AttributeError (#17949) 2026-01-29 09:35:12 -08:00
Shivam jindal
0769de9b0f Support LightOnOCR-2-1B (#17806) 2026-01-29 23:03:41 +08:00
Ziang Li
3c9cc44ff5 Add mxfp8 support for online quantization, Triton dense linear, and CUTLASS MoE (#17449) 2026-01-29 21:33:57 +08:00