chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity
9,304 Commits 1 Branch 0 Tags
3c9cc44ff5dad93f25cb41f4b448750fc0a32f34
Commit Graph

9304 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Ziang Li
3c9cc44ff5 Add mxfp8 support for online quantization, Triton dense linear, and CUTLASS MoE (#17449) 2026-01-29 21:33:57 +08:00
22dimensions
cfa09d311c [NPU] add support for npu x86_64 image release (#14127) 
Signed-off-by: 22dimensions <waitingwind@foxmail.com>
 2026-01-29 21:23:56 +08:00
Yuhao Yang
3c2f4c7bbe [diffusion] model: sync with upstream z-Image (#17822) 2026-01-29 21:10:11 +08:00
RoyWang
30adf78f82 [diffusion]: align sglang diffusion AMD pyproject_other.toml diffusion dependency with pyproject.toml (#16225) 
Co-authored-by: roywang <roywang@amd.com>
 2026-01-29 01:50:57 -08:00
kk
ef1c512754 Add aiter bias moe support in gpt-oss mxfp4 model (#17735) 
Co-authored-by: wunhuang <wunhuang@amd.com>
 2026-01-29 01:50:11 -08:00
triple-mu
319f6886fe [diffusion] model: move tp_rmsnorm check to WanTransformerBlock (#17792) 2026-01-29 16:39:00 +08:00
Zhang Yiyang (SII)
cdedbf1486 [diffusion] fix: resolve library mismatch in scheduler and update dit offload method name (#17916) 2026-01-29 15:54:36 +08:00
22dimensions
7b79326751 [NPU] support GPTQ quantization on npu (#15203) 
Signed-off-by: 22dimensions <waitingwind@foxmail.com>
 2026-01-29 15:48:18 +08:00
Niko Ma
cbf90d70ff [PD] Support KV transfer with MORI-IO (#14626) 
Co-authored-by: cwortman-amd <cwortman@amd.com>
 2026-01-28 23:22:41 -08:00
R0CKSTAR
d3cdee0a04 [MUSA][4/N] Add common device utilities, distributed backend, and custom op wiring (#17246) 
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
 2026-01-28 23:13:24 -08:00
Xinyuan Tong
9409c43593 Fix flaky tool calls in the Kimi K2.5 model (#17914) 
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
 2026-01-28 20:58:16 -08:00
Xinyuan Tong
3c34d2c3eb [FIX] kimi_k2 reasoning parser (#17901) 
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
 2026-01-28 19:47:09 -08:00
amote-i
1b22f2ee1c update ascend docs (#17741) 2026-01-29 11:43:48 +08:00
Joe Redmond
0ff0d181ca feat: add custom request header logging (#17786) 2026-01-28 19:33:08 -08:00
kk
f1384f5293 Integration mori backend for EP a2a data communication (#17012) 
Co-authored-by: Duyi-Wang <duyi.wang@amd.com>
Co-authored-by: billishyahao <bill.he@amd.com>
Co-authored-by: HaiShaw <hixiao@gmail.com>
 2026-01-28 19:07:34 -08:00
Jerry Ji
673dc09d9b [Fix][trtllm-mha] Canonicalize the strides when num_head = 1 (#17732) 2026-01-29 10:11:18 +08:00
Simo Lin
ac16d4450e [smg][mesh] extract mesh to mesh crate to reduce compile time (#17907) 2026-01-28 17:58:42 -08:00
Qi Yuhang
0368ddf9ea [JIT Kernel]Support fused_add_rmsnorm in JIT Kernel (#17677) 2026-01-29 09:29:59 +08:00
Zhang Yiyang (SII)
09a9147f59 [diffusion] model: support MOVA (#17704) 
Co-authored-by: gaoyang07 <Gary1546308416AL@gmail.com>
Co-authored-by: cms42 <c@cms42.top>
Co-authored-by: cms42 <44895820+cms42@users.noreply.github.com>
Co-authored-by: Ruixiao Li <cgruixiao@outlook.com>
Co-authored-by: Li Ruixiao(SII) <80368770+Li-dongyang@users.noreply.github.com>
 2026-01-29 09:12:08 +08:00
Prozac614
3fcda00e8c [CI] Fix CI timeouts by upgrading runai_model_streamer (related to #16937) (#17636) 2026-01-28 17:09:45 -08:00
Lianmin Zheng
d4180815a4 Make the functions in logits_processor.py and sampler.py more modular (#17885) 2026-01-28 16:24:23 -08:00
jackey hua
0998de088b [Perf] Tune Llama-4-Scout-17B-16E-Instruct fused moe kernel (#17891) 2026-01-28 14:06:46 -08:00
gingerXue
e9d727cb92 [MUSA][7/N] Enhance CUDA / PyNccl wrapper to support MTLink connectivity detection (#17499) 
Signed-off-by: jingzhi.xue <jingzhi.xue@mthreads.com>
Co-authored-by: jingzhi.xue <jingzhi.xue@mthreads.com>
 2026-01-28 11:36:30 -08:00
Артем Савкин
b77b0ffd60 [NPU] NZ for non-quantized MOE, Qwen3 MOE double memory consumption fix (#15904) 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
 2026-01-29 00:55:08 +08:00
Jinn
1953efb60e [AMD] ROCm: route W4A16 MoE to Triton and fix packed-weight loading (#17863) 2026-01-28 08:20:23 -08:00
triple-mu
1d1e72e516 [diffusion] fix: fix comfyui import typo (#17834) 2026-01-28 23:49:55 +08:00
Michael
f8636fbb25 [AMD] Add Kimi-K2, DeepSeek-V3.2 tests to nightly CI (#17523) 
Co-authored-by: YC Tseng <yctseng@amd.com>
 2026-01-28 00:55:46 -08:00
Even Zhou
6077de1237 [CI] [NPU] npu ci use existing modelscope model (#17868) 2026-01-28 16:51:41 +08:00
Xiaoyu Zhang
c08b54a575 [JIT kernel] Update jit_kernel cache and develop doc (#17842) 2026-01-28 15:09:47 +08:00
Mick
2573a262af [diffusion] doc: fix wrong docker run command (#17856) 2026-01-28 14:52:33 +08:00
Praneth Paruchuri
6f009961bb [model-gateway] Optimize HashRing construction to reduce heap allocations (#17575) 2026-01-27 22:46:40 -08:00
Praneth Paruchuri
897c35b457 [model-gateway] Optimize consistent hashing hot path to eliminate allocations (#17467) 2026-01-27 22:46:11 -08:00
Kangyan-Zhou
c0b4dd68a2 Add a performance dashboard server and frontend for nightly CUDA tests (#17725) 2026-01-27 22:22:33 -08:00
Xiaoyu Zhang
fb74e43707 [Diffusion] Delete sgl-kernel outdated time_embedding kernel (#17278) 2026-01-28 14:18:53 +08:00
Xiaoyu Zhang
67fb492c9a [CI] Fix test_moe_fused_gate error (#17844) 2026-01-28 12:03:17 +08:00
Ziang Li
a8dda2aa57 [DSv32] Overlap indexer qk projection and activation quant (#17688) 2026-01-28 11:46:49 +08:00
YC Tseng
52bca42870 [AMD] CI - enable deepseekv3.2 on MI325-8gpu and merge perf/accuracy test suites into stage-b suites (#17633) 
Co-authored-by: Bingxu Chen <Bingxu.Chen@amd.com>
 2026-01-27 18:54:36 -08:00
Yisheng Gong
1c4616a034 fix: add bias when enable mm fallback variant (#17690) 2026-01-28 09:50:49 +08:00
陈一涵
647428d8d6 [diffusion] perf: apply mul add fusion for Qwen-Image (#16299) 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
 2026-01-28 09:40:13 +08:00
Yashika Gandhi - Google
32ea7bcdd8 [diffusion] endpoint: fix vertex generate (#17611) 2026-01-28 09:38:56 +08:00
Mick
88fcd8535f [diffusion] feat: add an arg for controlling the number of prefetched layers in layerwise-offload (#17693) 2026-01-28 09:34:27 +08:00
Mick
1507dc6cdf [diffusion] fix: fix suppressing error log on non-main ranks (#17712) 2026-01-28 09:29:19 +08:00
Xiaoyu Zhang
331a22427c [Diffusion] glm-image apply flashinfer rope (#17689) 2026-01-28 08:51:37 +08:00
Hubert Lu
93423ff780 [AMD] Deprecate ROCm 6.3 artifacts and standardize gfx942 on ROCm 7 (#17785) 2026-01-27 15:58:49 -08:00
Liangsheng Yin
8278ef0e68 Pass GPU ids to kill specified devices in script. (#17840) 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
 2026-01-27 13:52:55 -08:00
siyu
4d00bd17a3 use shared memory for multimodal feature transport between Tokenizer and Scheduler (#16402) 
Co-authored-by: Yuhao Yang <47235274+yhyang201@users.noreply.github.com>
 2026-01-27 11:01:08 -08:00
Minglei Zhu
d90c0837e5 [hybrid-model] clean up and consolidate redundant fields in RadixLinearAttention (#17660) 2026-01-27 10:37:58 -08:00
Yi Zhong
8acd4d7d7e Make flashMLA work on: Cu13, B300 (#17600) 
Signed-off-by: vincentzed <207368749+vincentzed@users.noreply.github.com>
 2026-01-28 00:12:47 +08:00
fsygd
547e2d037e [diffusion] refactor: add arg to control the precision of dit (#17751) 2026-01-27 23:01:23 +08:00
Baizhou Zhang
1d942e4eef [DeepSeek] Update tests and document for DeepSeek V3.2 NVFP4 checkpoint (#17657) 2026-01-27 22:10:57 +08:00