Commit Graph

Select branches

Hide Pull Requests

bot/update-est-time-20260928

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• 863124684c [Spec v2] Remove allocate_lens and enable over-allocation (#13478) Liangsheng Yin 2025-11-22 22:49:10 +08:00
• 5625e32cae [VLM] Replace torch.repeat_interleave with faster np.repeat for Qwen-VL series (#13736) Yuan Luoandluoyuan.luo 2025-11-22 22:45:32 +08:00
• ca548d8324 [diffusion] refactor: refactor sampling params (#13706) Mick 2025-11-22 22:43:04 +08:00
• 3397bcee84 Tiny support different prompts in send_one.py (#13768) Liangsheng Yin 2025-11-22 21:59:58 +08:00
• 3e804bb0ba Update mem checker during busy (#13704) Liangsheng Yin 2025-11-22 21:15:02 +08:00
• a22de641ef [diffusion] fix: fix hunyuanvideo and add 2-gpu ci test (#13720) Yuhao YangandMick 2025-11-22 20:54:51 +08:00
• ac43822634 Refactor eagle bigram key matching (#13714) Liangsheng Yin 2025-11-22 20:40:42 +08:00
• a92afb00c6 [Ascend] support Kimi-K2-Thinking (#12759) 4 people 2025-11-22 17:33:07 +08:00
• 0eea17e3b5 Use dual stream for DS MoE whenever cuda graph is used (instead of with token threshold) (#9405) Trevor Morris 2025-11-22 08:37:41 +00:00
• 3805243245 modularize gsm8k and mmmu test classes (#13506) Netanel Haber 2025-11-22 10:06:53 +02:00
• 8bfce9b08d [Tiny] Renaming environ for NVFP4 dispatch (#13756) Baizhou Zhang 2025-11-22 00:05:20 -08:00
• b41afa3754 [Auto Sync] Update evict_policy.py, radix_cache.py (20251120) (#13669) 3 people 2025-11-21 22:07:36 -08:00
• 94ae816f29 Fix model weights validation with automatic cache cleanup (#13729) alisonshao 2025-11-21 19:40:00 -08:00
• 53620a1b1a fix test_lora_update.py starvation message check (#13702) Glen Liu 2025-11-21 22:33:04 -05:00
• 59b4d7f8d6 Fix B200 Nightly tests and move one manual test back to unit test to prevent the same issue (#13746) Kangyan-Zhou 2025-11-21 17:41:12 -08:00
• a56f770277 Fix global scaling factor loading hang (#13484) Shu Wang 2025-11-21 18:07:06 -06:00
• 1b48e1b974 Feat/nemotron nano v3 support (#12690) roikoren755 2025-11-21 23:53:05 +02:00
• a24aefe562 add some more labels (#13701) b8zhongandBrayden Zhong 2025-11-21 11:29:33 -08:00
• 45c572c58f Support torch 12.9 + DeepEP by removing custom nvshmem (#12949) fzyzcjyandBaizhou Zhang 2025-11-22 03:11:43 +08:00
• 681b9e6425 Revert "enable csgmv automatically on cuda" (#13707) Qiaolin Yu 2025-11-21 10:37:41 -08:00
• dab06b50ab Fix: CI monitor should not exit with error on regressions (#13694) alisonshao 2025-11-21 10:01:48 -08:00
• 85ffce30af [Piecewise CUDA Graph] Support Kimi-K2 (non-Thinking) (#13466) 3 people 2025-11-21 09:39:59 -08:00
• e94ef9fcae Add to ci permission (#13739) Jiajun Li 2025-11-21 12:21:09 -05:00
• 964cdedc50 [CI] minor hot fix of model validation list (#13737) Liangsheng Yin 2025-11-22 01:00:41 +08:00
• aa6e2c8aa0 [diffusion] log: minor improve logging (#13735) Mick 2025-11-21 23:01:23 +08:00
• 1776dce557 Super tiny delete unused files (#13734) fzyzcjy 2025-11-21 22:26:39 +08:00
• 99e13d189b Fix url: use https://roadmap.sglang.io for roadmap (#13733) Lianmin Zhengandsglang-bot 2025-11-21 05:42:32 -08:00
• dc836909a6 fix 3fs down, lock schedule main thread (#13407) weibingo 2025-11-21 19:02:36 +08:00
• 323fed5cfd [diffusion] doc: add contributing.md (#13649) Mick 2025-11-21 19:00:27 +08:00
• eff7df6d0a [AMD] Enable fused shared expert append and flatten quant for fp8 deepseekR1 model (#13705) jacky.chengandyctseng0211 2025-11-21 18:48:28 +08:00
• 5e7f91d451 [diffusion] profile: support performance metric dumping and comparison (#13630) Mick 2025-11-21 18:47:16 +08:00
• a34d3abb54 [Clean code] Compressed_tensors_moe code clean (#13719) Xiaoyu Zhang 2025-11-21 18:15:46 +08:00
• 6d0e0b9bfc [11/N] MoE Refactor: Simplifying SBO Implementation with Dispatcher Hooks (#13327) Cheng WanandBaizhou Zhang 2025-11-21 01:11:37 -08:00
• 589d9ad55b [NPU] chore: bump to CANN 8.3.RC1 and Pytorch 2.8.0 (#13647) Even Zhou 2025-11-21 17:07:08 +08:00
• a244c0309c [Fix] Qwen3Next lmhead dtype (#13708) Ziming Huang 2025-11-21 16:01:33 +08:00
• 1bb063aac8 [HiCache] fix unit test with changed new APIs (#13498) Teng Ma 2025-11-21 15:40:31 +08:00
• b30f63c40f [Bugfix] Fix hidden state size in EAGLE PD disaggregation buffers (#13590) 3 people 2025-11-21 08:32:40 +01:00
• 90a0133515 [Auto Sync] Update http_server.py, io_struct.py, scheduler_... (20251120) (#13679) 3 people 2025-11-20 23:17:15 -08:00
• 4360279036 [diffusion] server: use meta to avoid Linear init for TextEncoder (#13564) zyksirandMick 2025-11-20 23:14:24 -08:00
• b537ac0d1f Fix ZMQ bind error on non-zero rank nodes when using SGLANG_BLOCK_NONZERO_RANK_CHILDREN=0 (#13686) ishandhanani 2025-11-20 22:54:41 -08:00
• eda2f70033 [diffusion] doc: minor update docs (#13177) Mick 2025-11-21 14:35:29 +08:00
• 8c212a2029 [difusion] CI: speed up multimodal_gen ci (#13665) Yuhao YangandMick 2025-11-21 14:26:33 +08:00
• d754ce973e [Piecewise Cuda Graph] rename, refactor and add more logging (#13675) 4 people 2025-11-20 21:28:38 -08:00
• 475962a139 [VLM] Support Piecewise CUDA Graph for InternVL (#13640) Yuan Luoandluoyuan.luo 2025-11-21 13:21:51 +08:00
• bfcf15a129 [opt kimi k2 4 / n] Delete useless pad kernel in sgl_moe_align_block_size (#13587) Xiaoyu Zhang 2025-11-21 13:16:42 +08:00
• fb04d43428 [kimi k2 thinking] Avoid useless torch.zeros_ (#13596) 3 people 2025-11-21 13:15:27 +08:00
• 6be65ae462 Fix target MLA with eagle3 support for PD disaggregation (#13555) Mike QiuandMike_Qiu 2025-11-21 12:16:05 +08:00
• 750084ae08 remove unnecessary starvation check (#13619) Glen Liu 2025-11-20 22:10:51 -05:00
• 64480ec712 Add sgl-kernel CI test for Blackwell (B200) (#13301) alisonshao 2025-11-20 19:02:42 -08:00
• db2d362d04 [NVIDIA] Add cutedsl e2e test to GB200 CI (#12672) Kaixi HouandBaizhou Zhang 2025-11-20 18:30:12 -08:00
• 81e86992cd [CI] Move nightly tests to test/nightly/ (#13683) alisonshao 2025-11-20 18:00:02 -08:00
• c4db77f8a9 [model-gateway] fix gateway cli arg parser to not use = (#13685) Chang Su 2025-11-20 17:57:17 -08:00
• c0a2513b07 [diffusion] CI: improve validation method (#13627) Mick 2025-11-21 09:13:13 +08:00
• 3ae664d786 [model-gateway] add both python and rust cli alias (#13678) Simo Lin 2025-11-20 17:00:28 -08:00
• c56fc42430 Update quantization.md with new model resources (#13677) 赵晨阳 2025-11-20 15:50:16 -08:00
• 3f1cfd87b6 Super tiny remove unused MiniMaxM2MLP class (#13659) fzyzcjy 2025-11-21 07:35:12 +08:00
• b5344b31b8 [Piecewise CUDA Graph] Fix recompile issue for Mixtral and Grok2 (#13667) 4 people 2025-11-20 14:20:11 -08:00
• 6b262ac839 Test reorganization: Move tests to manual/ (#13610) alisonshao 2025-11-20 13:41:58 -08:00
• ada8ce1fd0 allow loras to be implicitly evicted and loaded based on max_loaded_loras (#11526) Glen Liu 2025-11-20 16:34:32 -05:00
• 5a2c70396e Add nightly test CI monitor workflow (#13038) alisonshao 2025-11-20 12:55:13 -08:00
• 42028af614 enable csgmv automatically on cuda (#13600) b8zhong 2025-11-20 12:53:02 -08:00
• 7291c72e57 [Deepseek V3.2] Change indexer weights_proj to fp32 (#13459) hlu1 2025-11-20 12:24:10 -08:00
• 6bc3062894 Fix launch of Olmo3 (#13666) Vincent Zhong 2025-11-20 15:04:30 -05:00
• fa92441027 [DeepseekV3.2] Deepseek fp8 support for MHA path (#12964) YAMY 2025-11-20 11:13:36 -08:00
• fc9efdcb98 Adding nightly tests as release guard for bot bump workflows (#13655) Douglas Yang 2025-11-20 10:04:36 -08:00
• 2dec555d36 [10/n] decouple quantization impl from vllm dependency - fix import (#13524) Fan Yin 2025-11-21 01:45:51 +08:00
• acde21d8d5 Add fused_rmsnorm_gated_cpu kernel for CPU to support Qwen3-Next (#11577) YanbingJiang 2025-11-21 01:33:31 +08:00
• 4528cb7d40 [CI] apply pr-gate for XPU (#13663) Liangsheng Yin 2025-11-20 23:07:04 +08:00
• a352e833c4 CI: Kill zombie diffusion processes in CI & minor code style fix on rotary embedding fallback (#13637) Lianmin Zheng 2025-11-20 05:57:01 -08:00
• 852eb6ce2a [CI] optimize CI workflow info (#13634) Liangsheng Yin 2025-11-20 18:12:51 +08:00
• 7af9b88c6c Revert "[Feature] Introduce JIT Kernel in sglang (with hicache JIT kernel)" (#13644) Lianmin Zheng 2025-11-20 02:11:12 -08:00
• 2847e5c4b4 fix bench_speculative bug (#13197) Lzhang-hub 2025-11-20 17:09:04 +08:00
• c8ede0e93c [ROCM] Optimized deepseek-r1 fp8 model with + triton_gemm_a8w8 + batch_gemm_a8w8 + fused set_mla_kv_buffer kernel (#13617) 3 people 2025-11-20 16:29:56 +08:00
• 19729f723e [CI] Align metric units for CI rate limit (#13633) Liangsheng Yinandgemini-code-assist[bot] 2025-11-20 16:25:57 +08:00
• b51f9bbee7 [Feature] Introduce JIT Kernel in sglang (with hicache JIT kernel) (#13453) DarkSharpness 2025-11-20 16:03:32 +08:00
• 4a8442af1b diffusion: improve baseline performance monitor (#13614) Mick 2025-11-20 14:29:56 +08:00
• 7dcf910d53 Add support for new aiter version (AR accuracy, is_shuffled PR) (#13554) Thomas Wangandsogalin 2025-11-20 14:17:57 +08:00
• c7b37b7074 [diffusion] fix: remove multimodal_gen redundant get_bool_env_var func (#13583) joesunandMick 2025-11-20 14:05:29 +08:00
• 10e0b83a4c Add FP32 dtype support for RoPE - Part2 (#13328) iLeGend 2025-11-20 13:19:53 +08:00
• bc42c8c415 [diffusion] refactor: refactor pipeline folders (#13253) Mick 2025-11-20 12:56:51 +08:00
• 2e3a69ae05 [diffusion] refactor: remove PreprocessorConfig (#13248) Mick 2025-11-20 12:53:26 +08:00
• 21370ef7b5 [UT] Destroy process group after broadcast to resolve port occupation issues in multi-server tests (#12379) Zeyu Li 2025-11-20 12:51:58 +08:00
• c3c4da71fb [NVIDIA] Add fp8 gemm benchmark on blackwell (#13528) Kaixi Hou 2025-11-19 19:35:00 -08:00
• dc69462456 [CI fix] Fix image download failures in VLM CI tests (#13613) Xiaoyu Zhang 2025-11-20 11:18:06 +08:00
• 48ca9f7518 feat: support external custom models (#13429) 3 people 2025-11-20 11:16:58 +08:00
• 127d59cd2c [diffusion] CI: improve diffusion CI (#13562) MickandAdarsh Shirawalmath 2025-11-20 10:54:13 +08:00
• af6bcadcf7 [VLM] Support Piecewise CUDA Graph for Qwen2.5-VL (#13055) 3 people 2025-11-20 10:23:44 +08:00
• 67fca6b297 [GDN] Remove unnecessary conv state clone (#13603) Binyao Jiang 2025-11-19 16:09:12 -08:00
• f88b2aa6af [GDN] Remove unnecessary contiguous() (#13604) Binyao Jiang 2025-11-19 16:07:49 -08:00
• 17b24aca6e [Auto Sync] Update base_grammar_backend.py, collector.py (20251116) (#13357) 3 people 2025-11-19 16:05:50 -08:00
• bfaf0b8607 chore: bump sgl-kernel version to 0.3.17.post2 (#13570) sglang-bot 2025-11-19 14:02:57 -08:00
• 83756a4b33 add https://github.com/netanel-haber to CI_PERMISSIONS.json (#13577) Netanel Haber 2025-11-19 23:20:30 +02:00
• a355794905 Expend compatibility check for all quantized MoE models (#13465) Xinyuan Tong 2025-11-19 17:24:27 +00:00
• e72cf13693 Support moe topk sigmoid kernel (#13049) Roger Youngandxuebi 2025-11-20 00:24:37 +08:00
• 196b940aed [3/N] CI refactor: move some manually triggered tests. (#13448) Liangsheng Yin 2025-11-19 23:06:53 +08:00
• a1e1e533b9 Tiny enhance test suites sanity check (#13589) Liangsheng Yin 2025-11-19 22:43:58 +08:00
• d4a4dcdfb3 [NPU] Adapt pr-gate for pr-test workflow & workflows refresh (#13567) Even Zhou 2025-11-19 21:55:17 +08:00
• 97ba2c2d1e [CI] Fix CUDA workflow's dependency. (#13568) Liangsheng Yin 2025-11-19 19:06:42 +08:00
• e197bef5ce [RL] Allow passing tensors of different dtypes for FlattenedTensorBucket (#13413) Zilin Zhu 2025-11-19 17:24:48 +08:00
• 8900f996aa CI Failure Monitor Improvements (#13558) Douglas Yang 2025-11-18 22:55:38 -08:00