sglang

minke.yu/sglang

Fork 0

0a1fb0da86 add LRU eviction for mooncacke embedding cache (#25954)
Mike Qiu Mike_Qiu siyu Yuang Chen
2026-06-12 12:38:13 +08:00
e1164a6dfc [Spec] Remove dead prepare_for_verify / prepare_extend_after_decode + extend-decode kernel (#27761) Liangsheng Yin 2026-06-11 20:59:30 -07:00
3a3a759464 [NPU] Add Gemma4 Sliding Window Attention support on Ascend backend (#26147) syy-hw 2026-06-12 11:44:38 +08:00
8bfcc0c39c Use the correct wrapper for fp4_quantize (#27956) Brayden ZhongandBrayden Zhong 2026-06-11 20:23:09 -07:00
40894be3c3 add lfm2.5 to new cookbook. (#27409) Yi Zhong 2026-06-11 20:08:19 -07:00
f5c9f88ee2 [plugin][distributed] use active platform's backend in get_default_distributed_backend (#23969) Yihao Wang 2026-06-11 20:05:17 -07:00
7074704c0c [Spec] Dedup post-verify mamba state commit into shared spec_utils helpers (#27966)
Liangsheng Yin xbfs Cursor
2026-06-11 18:33:40 -07:00
3ffe72517f [Spec] Remove the dead spec V1 scheduler paths (#27977) Liangsheng Yin 2026-06-11 18:31:13 -07:00
2e74ff192c [DSV4] Use int64 for compressor out_loc tensors (#27973) Cheng Wan 2026-06-11 17:45:34 -07:00
97a0031799 [lint] Enable Ruff UP037 to drop redundant quoted annotations (#27984) Cheng Wan 2026-06-11 17:38:05 -07:00
c0480a88be [Spec] Retire Spec V1 (#27964) Liangsheng Yin 2026-06-11 16:15:15 -07:00
949326d922 Add SGLANG_ENABLE_WAR_BARRIER to force-enable the overlap scheduler WAR barrier on non-CUDA (e.g. AMD) (#27967) Oguz Ulgen 2026-06-11 15:38:37 -07:00
493f828bfa Add DeepGEMM prerelease wheel tests (#27075) Brayden ZhongandBrayden Zhong 2026-06-11 15:14:50 -07:00
d71e9bede6 [bugfix] commit Mamba states after NGRAM target verify (#26351)
Bofeng Xue xbfs Cursor
2026-06-12 06:11:18 +08:00
cd075d1f64 [RL] convert DeepSeek V4 APE layout through weight loader (#27307) Yueming Yuan 2026-06-11 15:05:27 -07:00
0bac184425 [NVIDIA] Update Minimax-M2.5,M2.7 docs with flags for performance (#24465) Trevor Morris 2026-06-11 14:58:44 -07:00
fee717f303 [Spec] Fold the DFLASH worker base into DFlashWorkerV2 on BaseSpecWorker (#27950) Liangsheng Yin 2026-06-11 14:54:28 -07:00
acdb39edd2 [Spec] Remove the DFLASH V1 worker path (#27959) Liangsheng Yin 2026-06-11 14:37:58 -07:00
6e885c844f Revert "[AMD] Fix DeepSeek V4 Pro c128 state tensor dtype mismatch error and c4_sparse_raw_indices attribute error in cuda graph phase" (#27919) Wang, FangYuan 2026-06-12 05:25:32 +08:00
12d3f02be8 [AMD] Fix DSA device-to-host direct test on rocm720 (page_size%16 assert) (#27850) Michael 2026-06-11 14:24:23 -07:00
df5055e00f Bump spec logprob match delta for the bf16 eagle fixture (#27952) Liangsheng Yin 2026-06-11 14:21:07 -07:00
1ac75c3463 [CI] Fix GB300 TestDummyWithSBO crash: disable NaN assert and coredump for dummy weights (#27955) Liangsheng Yin 2026-06-11 14:17:03 -07:00
d26b90cf70 Add weireweire to CI permissions (#27957) Baizhou Zhang 2026-06-11 14:10:45 -07:00
ec0eb6cce8 Support MiMo v2 ASR (#26278)
Qingchao Zhu github-actions[bot]yanyihan zhuqingchao
2026-06-12 04:43:13 +08:00
24c5d76f74 fix: per-sequence last-token embedding in EAGLE3/MTP draft for batched multimodal spec decoding (#27846) Jackey Hua 2026-06-11 13:33:16 -07:00
10219bd9d6 [PD] Fix negative prefill kv_transfer_alloc_ms under optimistic prefill (#27885) cctryandcctry 2026-06-11 13:12:25 -07:00
880e6f66fc [BCG] Share output buffers across capture sizes + typed ShapeKey (#27857) Yuwei An 2026-06-11 11:58:05 -07:00
7f57b344c9 [diffusion] feat: progressive resolution growing for Ideogram 4 via GPU DCT upsampling with up to 1.56× speedup (#27736) Brian Chao 2026-06-11 23:16:53 +08:00
b2728bda9d [diffusion] feat: use fused w8a8 kernel for Ideogram4 weight-only linear as an opt-in (#27590) Chi McIsaacandMick 2026-06-11 11:15:27 -04:00
06e0df5899 Optimize Qwen3 Next FP8 MoE on H200 (#26204) Xiaoyu ZhangandBBuf 2026-06-11 22:18:43 +08:00
1a6b5561db Fix MLA scaling when YARN scaling is disabled (#26203) Xiaoyu ZhangandBBuf 2026-06-11 22:17:55 +08:00
d571e076fa [codex] Centralize more inline Triton kernels (#27429) Xiaoyu Zhang 2026-06-11 22:17:26 +08:00
d9110d971e [diffusion] fix: fix wan ti2v sp timestep padding (#27876) Mick 2026-06-11 22:13:25 +08:00
8077fb1df7 fix(deepgemm): align PP-parallel warmup bs to CP padding (#27922) ybyang 2026-06-11 20:52:45 +08:00
9e9fde1478 [diffusion] Revert "Mistral3 add tensor parallel support for diffusion text encoder " (#27892) Mick 2026-06-11 20:33:45 +08:00
6a012fbb2d [PD] Fix ZMQ stale socket reconnection in PD disaggregation (#27796) Shangming CaiandAbatom 2026-06-11 19:51:12 +08:00
66076f2409 [mem_cache][3/N] refactor: move HiSparse allocators to allocator/hisparse.py (#26678) shuwennandClaude Opus 4.8 2026-06-11 19:50:24 +08:00
be45745f38 [EPD] fix: zmq PUSH socket reconnect-aware connection management with tcp keepalive (#27039) Zhonghua Deng 2026-06-11 19:32:31 +08:00
7e245afefe [CI] Fix registered sigmoid gate mul test location (#27909) Mohammad Miadh Angkad 2026-06-11 18:52:37 +08:00
6ac9f66596 Remove MoE prefill CUDA graph disable guard (#27841) Mohammad Miadh Angkad 2026-06-11 18:30:15 +08:00
54cba63b6c Fix paged SWA free mapping cleanup (#27779) luoroger37 2026-06-11 18:25:24 +08:00
22c7285a26 [AMD] Fuse sigmoid + mul attention output gate into single Triton kernel (#27630) jacky.cheng 2026-06-11 17:05:05 +08:00
f4b3b99413 [AMD] Enable fused GDN QKV split Triton kernel on HIP (#27583) jacky.cheng 2026-06-11 17:01:27 +08:00
493bb6ae0d Fix fp16 NaN flake in spec CI: bf16 eagle fixture; sanitize NaN logits in sampler (#27883) Liangsheng Yin 2026-06-11 01:16:12 -07:00
b8376aebd0 [AMD] Fix the dsv4 performance of MoE issue. (#27858) kkandwunhuang 2026-06-11 16:06:54 +08:00
2a51479a91 ci(xpu): pull intel/sglang-dev:latest and clean workspace properly (#27860) ashwini rathi 2026-06-11 10:44:56 +05:30
57907cff97 docker(xeon): support running container as non-root user (#27157) Vinay Kumar 2026-06-11 09:56:26 +05:30
dc1e46ec8f [Intel GPU]Add sycl mrope pass for xpu device (#27646) gaopengff 2026-06-11 12:25:55 +08:00
9425ba2478 Remove extra_pytest_path from pr-test.yml (#27856) Cheng Wan 2026-06-10 21:16:14 -07:00
5e0271536a [UnifiedTree]: HybridModel launches HiCache via UnifiedTree by default. (#27759) Zhangheng 2026-06-11 12:03:36 +08:00
9788c8e867 [RL] Handle Mooncake buffers across memory release (#27696) Zilin Zhu 2026-06-11 10:15:00 +07:00
f8b0a120b8 fix: DSV4 BCG compress-prefill plan OOB on underfilled (tiny) prefill replay (#27747) YAMY 2026-06-10 20:02:24 -07:00
f4f30d7d23 [Fix] Use int64 seq_lens across all CUDA graph runners and backends (#27840) Cheng Wan 2026-06-10 19:54:58 -07:00
b4bed8c398 [diffusion] fix: cast to float32 (from float64) in triton kernel to unblock torch.compile (#27709) Yihao Wang 2026-06-10 19:29:21 -07:00
588d1f7bc9 [Feature] Spec V2 DFlash Support (#23000) David Wang 2026-06-10 22:27:42 -04:00
ce0ff154a5 add mimo best practice (#27665) ming_wang 2026-06-11 10:15:31 +08:00
475e9d25bf bugfix for npu mtp graph runner (#27808) Hexq0210 2026-06-11 10:09:50 +08:00
fc1fee528c [NPU] [DOC] replace <code> with backticks and remove obsolete params (#27677) amote-i 2026-06-11 09:33:31 +08:00
43835b5ba5 docs: cookbook benchmark accuracy labels come from the model config (no engine default) (#27842) zijiexiaandClaude Fable 5 2026-06-10 18:27:38 -07:00
66989a7642 Fix gpt-oss-20b with mxfp4 support for Xeon (#27782) YanbingJiang 2026-06-11 09:16:16 +08:00
e113d884d0 ci(xpu): checkout deadicated dir for night build (#27703) Jitendra Patil 2026-06-10 20:09:46 -05:00
99ab90c5b7 [AMD] Restore AMD piecewise CUDA graph support dropped by #23906 (#27811) fxmarty-amd 2026-06-11 02:43:48 +02:00
5f913c1135 [Fix] Emulate PDEATHSIG on macOS to prevent orphaned worker processes (#27190) Lijuan TangandClaude Opus 4.8 2026-06-10 17:43:27 -07:00
db061e97c0 [Unified] Fix UnifiedRadixCache write_backup issue in write-back mode(#27108) Yongji Wuandhzh0425 2026-06-10 17:28:24 -07:00
740305e1d9 [HiCache] Add opt-in LRU eviction to file storage backend (CP-aware) (#26670)
HZY Zhangheng Claude Opus 4.8
2026-06-11 08:21:59 +08:00
125ef88892 Disable async assert in Nemotron nightly tests (#27838) Baizhou Zhang 2026-06-10 16:28:14 -07:00
16124fc9b2 [Metrics] Fix fwd_occupancy reading NaN on every decode log line; probe-free base-a (#27836) Liangsheng Yin 2026-06-10 15:42:19 -07:00
bdf3ef6421 [CI] Fix registered QK Gemma RMSNorm test location (#27839) Mohammad Miadh Angkad 2026-06-11 06:21:59 +08:00
0da18f8d91 [AMD][Perf] Fuse QK RMSNorm + gate extraction Triton kernel for Qwen3.5 on HIP (#27656) jacky.cheng 2026-06-11 05:29:33 +08:00
3c1b0fb226 [1/n] [CP] Simplify prefill context parallel server args (#27312) Baizhou Zhang 2026-06-10 14:11:38 -07:00
73d0989d9b docs: make playground issue template model field a free-form input (#27827) zijiexiaandClaude Fable 5 2026-06-10 14:04:16 -07:00
5fefe91289 [Spec] NGRAMWorker on BaseSpecWorker; algo-owned verify-tree shape params (#27799) Liangsheng Yin 2026-06-10 12:38:16 -07:00
99258b2f1e [Docs] Restore right-hand ToC on the DeepSeek-V4 cookbook page (#27830) zijiexiaandClaude Opus 4.8 2026-06-10 11:58:02 -07:00
3600a9ac5f [SPEC] feat: init adaptive spec params from config (#27493) shuwennandQiaolin Yu 2026-06-11 02:55:25 +08:00
21647f1f5d [router] Apply chat template before cache-aware hashing (fix overlap=0 on chat traffic) (#27386) Kangyan-ZhouandClaude Opus 4.8 2026-06-10 10:27:54 -07:00
276c98c6cf [Docs] Add Kimi-K2.6 NVFP4 and update Kimi-K2.5 cookbook guidance (#27714) Mohammad Miadh Angkad 2026-06-11 01:24:09 +08:00
0ae27405d0 [AMD] Support eplb for moriep (#22985) billishyahaoandHAI 2026-06-11 01:23:51 +08:00
91ff7baa28 [Docs] Add GLM-5.1 NVFP4 to cookbook (#27708) Mohammad Miadh Angkad 2026-06-11 01:23:15 +08:00
8c6bbe0658 [deepseek] Enable DP attention + TBO + shared experts fusion (#27510) Kai-Hsun Chen 2026-06-10 09:42:27 -07:00
1cf8efdd08 docs: Diffusion Gemma cookbook (#27824)
Khoa Pham Richard Wang Claude Opus 4.8
2026-06-10 09:33:01 -07:00
0c7faf01fb test(sgl-router): cover sticky scale-up no-redistribution e2e (#27742) Kangyan-ZhouandClaude Opus 4.8 2026-06-10 08:54:31 -07:00
6a16f29af6 [AMD] ci: register 8 framework / unit tests to run on AMD CI (#25939) Michael 2026-06-10 08:41:22 -07:00
502bc89e1b [AMD] Fix DeepSeek V4 Pro c128 state tensor dtype mismatch error and c4_sparse_raw_indices attribute error in cuda graph phase (#27529)
Wang, FangYuan YC Yen-Ching Tseng Thomas Wang
2026-06-10 23:30:48 +08:00
53ed34cb88 Fix MiMo-V2.5-Pro DP-attention dp size in cookbook deployment snippet (#27668) JoyFuture 2026-06-10 22:08:28 +08:00
518e35fae7 [KDA] Add CuteDSL Prefill Kernel on SM100 (#27488) Yuan Luoandluoyuan.luo 2026-06-10 21:25:19 +08:00
4faaa9ba92 [CI] Fix stale ngram bookkeeping owner sites (#27803) Mohammad Miadh Angkad 2026-06-10 19:51:46 +08:00
b0d888a195 [CI] Remove AMD DSv4 Docker publish job (#27795) Thomas Wangandbingxche 2026-06-10 17:50:28 +08:00
111009ea54 [Feature] [Ngram spec] Support ngram spec v2 (#17260)
Siyuan Chen hnyls2002 Ratish P
2026-06-10 17:46:00 +08:00
255843d454 Support for Zyphra zaya1 model (#26347) ChengYao-amdandgithub-actions[bot] 2026-06-10 17:44:47 +08:00
b40f365732 [CI] Move misplaced mhc kernel test into test/registered/kernels (#27781) DarkSharpnessandClaude 2026-06-10 16:49:18 +08:00
70c71ba183 [NPU] Fix dead patch_model monkey-patch breaking NPU torch.compile capture (#27774) Cheng Wan 2026-06-10 01:15:50 -07:00
01f10acd06 Implement online nvfp4 quantization (#26083) Ziang Li 2026-06-10 00:26:51 -07:00
e76e4959b5 [CI][PD] Add unit tests for nixl backend (#26908) nbarzilie 2026-06-10 09:31:59 +03:00
1a5775a9df [Docs] Remove the legacy release-docs.yml deploy workflow (#27766) zijiexiaandClaude Opus 4.8 2026-06-09 23:18:41 -07:00
e8a437ef26 [diffusion] doc: update docs architecture (#27767) Mick 2026-06-10 14:18:10 +08:00
f2bcdb0508 [AMD] Add unified kv attention support in dpsk-v4 (#27380) Thomas WangandXinyi Song 2026-06-10 14:13:37 +08:00
95d8a75bc9 Bundle set_kv_buffer write targets into KVWriteLoc (loc + swa_loc) (#27695) Cheng Wan 2026-06-09 23:09:51 -07:00
758fd4bb9a [SWA] Cache full→SWA out_cache_loc per forward across attention backends (#27617) Cheng Wan 2026-06-09 22:57:51 -07:00
08ceb96ea5 [diffusion] fix: remove boolean arithmetic guard to fix compiling (#27065) Aleksi Vesanto 2026-06-10 08:55:47 +03:00
4704b10d0d [AMD] Update MoRI to v1.2.0 (#27669) Bingxu Chen 2026-06-10 13:41:53 +08:00

... 47 48 49 50 51 ...

Commit Graph Select branches Hide Pull Requests dsv41-pd dsv41-pd-visioncp main Mono Color

Commit Graph

Select branches

Hide Pull Requests

dsv41-pd

dsv41-pd-visioncp

main