Commit Graph

4786 Commits

Author SHA1 Message Date
Douglas Yang
697a77bf73 Add stress test workflow (#13937)
Co-authored-by: Liangsheng Yin <hnyls2002@gmail.com>
2025-11-26 13:24:39 -08:00
Lianmin Zheng
0a186924ba [Code sync] Fix registration of some ops in grok & Fix oss sync scripts (#13990)
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
2025-11-26 13:11:52 -08:00
Baizhou Zhang
8a9b8b8457 Revert "Fix nightly test failures: NSA indexer dtype and CPP radix cache init" (#14015) 2025-11-26 10:45:23 -08:00
Netanel Haber
8308cd3632 Support internvl on Blackwell (which doesn't support fa3): add SingletonCache support to Vision{Sdpa|Triton|Ascend}Attention (#13151) 2025-11-26 10:31:45 -08:00
Da Chen
e0e8a99630 fix: correct usage of minimax-m2 deepep moe forward (#13892)
Co-authored-by: Dash <dash@minimaxi.com>
Co-authored-by: 赵晨阳 <zhaochen20@outlook.com>
2025-11-26 10:01:23 -08:00
Mick
5102d00901 [diffusion] model: support black-forest-labs/FLUX.2-dev (#14000) 2025-11-27 01:49:40 +08:00
Vedant V Jhaveri
9dab534b35 fix spec dec request level metrics (#13754)
Co-authored-by: Liangsheng Yin <lsyincs@gmail.com>
2025-11-27 01:09:21 +08:00
Michelle Wu
262c3c1fde [Ascend] Support enable-mixed-chunk in non-MLA scenarios (#12491)
Co-authored-by: ronnie_zheng <zl19940307@163.com>
2025-11-27 00:00:21 +08:00
Liangsheng Yin
6c190cbda0 Rename: --hooks to --forward-hooks (#13994) 2025-11-26 22:26:28 +08:00
Liangsheng Yin
eff6a07c8f Fix get_load API (#13991) 2025-11-26 22:07:51 +08:00
Shangming Cai
b704b0a922 Optimize uneven PP layer distribution logic to improve PP performance (#13977) 2025-11-26 22:05:03 +08:00
vipwangerxiao
15729dbc8e Use dynamically maintained num_waiting_tokens in get_load() (#13203)
Signed-off-by: Peng Wang <peng_wang@linux.alibaba.com>
Co-authored-by: Peng Wang <peng_wang@linux.alibaba.com>
2025-11-26 20:34:45 +08:00
Zehuan Li
21b0582d4b [feature] Initial block diffusion language model support (#12588)
Co-authored-by: Tiwei Bie <tiwei.btw@antgroup.com>
2025-11-26 17:57:54 +08:00
Yuhao Yang
5795da5e83 [diffusion] fix: fix the issue where the qwen-edit & wan model produces incorrect output during sequence parallelism (#13922) 2025-11-26 15:58:23 +08:00
StonyPort
540d6fee20 Support piecewise CUDA graph for embedding models (#13852)
Co-authored-by: qiuxuan.lzw <qiuxuan.lzw@alibaba-inc.com>
2025-11-26 15:29:46 +08:00
ShawnY112358
007c3e234c [feat] support in-flight weight update (#10071)
Co-authored-by: 赵晨阳 <zhaochen20@outlook.com>
2025-11-25 22:03:13 -08:00
fzyzcjy
7130ad3a29 Fix SGLANG_ENABLE_HEALTH_ENDPOINT_GENERATION not working (#13961) 2025-11-25 21:58:55 -08:00
alisonshao
846ba3c62c Fix nightly test failures: NSA indexer dtype and CPP radix cache init (#13958) 2025-11-26 11:52:08 +08:00
Yubo Wang
18fb51583f Support FlashAttention3 page_size > 1 and topk > 1 case with paged attn and spec decode (#7725) 2025-11-26 11:44:41 +08:00
Yuan Luo
ca5c8b16f6 [VLM] Support InternVL Vision Encoder Data Parallelism (#13925)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2025-11-26 11:43:05 +08:00
timmy-feng
f33e5d1ef1 fix: spec overlap predict shape does not match verify output shapes (#12786) 2025-11-26 11:01:06 +08:00
Kangyan-Zhou
13e5beeab4 Fix Deepseek v3.1 loading issue (#13954) 2025-11-25 18:28:52 -08:00
sglang-bot
c53e729d45 chore: bump sgl-kernel version to 0.3.18.post1 (#13951)
Co-authored-by: sglang-bot <sglang-bot@users.noreply.github.com>
2025-11-25 18:14:28 -08:00
Fan Yin
36b1bcd242 [chore] update torch version to 2.9 (#12969)
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
2025-11-25 14:47:34 -08:00
fzyzcjy
64a11303ce Fix update weight error for blackwell DeepGEMM (#13910) 2025-11-25 13:28:12 -08:00
Lianmin Zheng
1ab6ce0e62 [Auto Sync] Improve profilers and simplify bench_one_batch_server.py (#13866) 2025-11-25 12:13:31 -08:00
Jan Bernlöhr
fcccaf9001 Add Llama4 attention backend auto-selection (#13421)
Signed-off-by: jbernloehr <jbernloehr@nvidia.com>
2025-11-25 11:54:21 -08:00
Tova Movshovitz
215a97fa6c Fix docstrings for v1 HiCacheStorage methods (#13851) 2025-11-25 11:48:26 -08:00
YAMY
5eed5fc0b0 [DeepSeekV3.2] Centralize NSA dispatch logic in NativeSparseAttnBackend (#13544)
Co-authored-by: hlu1 <14827759+hlu1@users.noreply.github.com>
2025-11-25 11:32:30 -08:00
Mick
dfd7ab9682 [diffusion] feat: support LoRA (#13859) 2025-11-26 00:21:33 +08:00
Mick
46673b4224 [diffusion] doc: add doc for LoRA usage (#13931) 2025-11-26 00:02:14 +08:00
Liangsheng Yin
d3d404d3d7 [CI] CI registry update (#13927)
Co-authored-by: alisonshao <54658187+alisonshao@users.noreply.github.com>
2025-11-25 22:43:19 +08:00
luchangli
64225a8ae9 fix nixl prefill crash make decode health check failed (#13657)
Signed-off-by: liluchang <liluchang@kingsoft.com>
2025-11-25 21:40:11 +08:00
Chen1022
d64bf6c6ce Support piecewise cuda graph for Qwen3-next (#13081) 2025-11-25 21:01:27 +08:00
Liwansi
432ecf841e [Ascend] qwen optimization (#12078)
Co-authored-by: Even Zhou <even.y.zhou@outlook.com>
2025-11-25 19:44:24 +08:00
Mick
6f094deff0 [diffusion] CI: minor refactor CI for less code duplication (#13905) 2025-11-25 18:44:11 +08:00
ant-yy
59464dbf15 [Fix]: Further fix the buffer len of future map (#13916)
Signed-off-by: vito.yy <vito.yy@antgroup.com>
2025-11-25 18:09:28 +08:00
Yi Zhang
1f7fcc10d5 [diffusion] profile: fix profiling bugs (#13642)
Co-authored-by: Mick <mickjagger19@icloud.com>
2025-11-25 17:58:35 +08:00
Lzhang-hub
760c20b360 update flashinfer_cubin==0.5.3 (#13848) 2025-11-25 00:10:34 -08:00
DarkSharpness
c1dd9a9599 [Fix] JIT kernel dependencies in other platforms (#13889) 2025-11-24 23:19:17 -08:00
gongwei-130
8ff3ef1fef fix: draft model revision misuse model revision (#11893) 2025-11-24 21:13:37 -08:00
Yibo Cai
da182e4b83 [CI] fix lint error (#13891) 2025-11-24 20:51:33 -08:00
alisonshao
83e7207763 [diffusion] CI: add validation and cleanup for corrupted safetensors in multimodal loader (#13870)
Co-authored-by: Mick <mickjagger19@icloud.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2025-11-25 12:02:43 +08:00
Mick
9384fa2729 [diffusion] refactor: remove training-related code (#13860) 2025-11-25 11:38:50 +08:00
Nicolas Castet
b0a26ba624 Add support for bf16 x bf16 cutlass fused MoE (#10275)
Co-authored-by: Sam Li <lsam@nvidia.com>
Co-authored-by: jackeyhua <jackeyhuasjtu@gmail.com>
2025-11-24 18:49:39 -08:00
Binyao Jiang
de430b6745 [Performance] Replace preprocess_video logic from GLM multimodal processor with transformer impl for speed up (up to 27% faster) and addressing OOM (up to 50x improvements) (#13487) 2025-11-24 18:17:13 -08:00
Qiaolin Yu
4b45d556a7 Overlap glm moe gemms in two cuda streams (#13786) 2025-11-24 18:15:24 -08:00
Even Zhou
db0ffc09ef [NPU] Fix NPU CI (#13834)
Co-authored-by: c30031083 <chenxu140@huawei.com>
2025-11-25 10:09:36 +08:00
Glen Liu
eb1d885400 add LoRA warning if loading a preexisting LoRA adapter with a different name (#13822) 2025-11-24 15:16:41 -08:00
Lianmin Zheng
e83bd1fadc [Auto Sync] Update schedule_batch.py, schedule_policy.py, b... (20251122) (#13763)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Hanming Lu <69857889+hanming-lu@users.noreply.github.com>
Co-authored-by: Hanming Lu <hanming@x.ai>
2025-11-24 14:33:31 -08:00