Commit Graph

Select branches

Hide Pull Requests

bot/update-est-time-20260928

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• 4a78031a71 [ROCM] Optimized deepseek-r1 model with rmsnorm + fp8 quant fusion (#12689) yctseng0211 2025-11-11 18:59:10 +08:00
• ea10a9d165 [bug][rocm]fix qr when variable inp (#11609) haoyangli-amd 2025-11-11 17:43:48 +08:00
• 71aea45c41 [Fix] Add TPOT back to bench_serving (#12976) elvischenv 2025-11-11 17:32:24 +08:00
• e3b38d7194 [Bug] TypeError: maybe_executor_submit() (#13050) Johnsonms 2025-11-11 00:55:13 -08:00
• 5c0cadd0c6 Remove duplicate import (#12980) LHXuuu 2025-11-11 16:53:39 +08:00
• 39b1d048a0 [AMD] Add PD test for AMD CI (#11938) michael-amd 2025-11-11 00:49:44 -08:00
• 8db7fc4186 disable overlap schedule if mamba radix cache open (#13057) Yi Zhang 2025-11-11 16:35:38 +08:00
• fe92d4d88e [CI] Auto format code (#13053) Xiaoyu Zhang 2025-11-11 15:30:29 +08:00
• fc8cda14cf Sglang Tracing: optimize trace_event_batch() (#13036) Feng Su 2025-11-11 15:02:20 +08:00
• 6a7322ffbc [diffusion] doc: add support_new_models.md (#13043) Mick 2025-11-11 12:49:48 +08:00
• c751cb38b0 [AMD CI] Update CI Version Logic. (#13029) Sai Enduri 2025-11-10 20:42:00 -08:00
• 3594815a8b Re-enable Flashinfer TRTLLM GEN MHA and Add Unit Test (#12885) Sam 2025-11-11 12:17:43 +08:00
• 9caca6a45c [PieceWise CUDA Graph] Support awq/gptq model in piecewise cudagraph (#12518) Xiaoyu Zhang 2025-11-11 11:56:15 +08:00
• 08c805a85f fix(ci): workflow id in permission rate limit (#13035) yinghui 2025-11-10 19:06:13 -08:00
• f18ec927f3 fix tuning_fused_moe_triton_sep tool per_channel_quant bug (#13027) Xiaoyu Zhang 2025-11-11 10:33:54 +08:00
• aea88fa7af [AMD CI] Update docker release workflows docker file name. (#13028) Sai Enduri 2025-11-10 18:28:39 -08:00
• 2fe4e69fca [router] add postgres databases data connector (#12218) rongfu.leng 2025-11-11 08:51:50 +08:00
• 012bfc4fdc [9/n] decouple quantization impl from vllm dependency - adjust ci (#12753) Peng Zhang 2025-11-11 06:55:19 +08:00
• 0493775b06 [router][ci] Quick Improvement to make CI more stable (#12869) Keyang Ru 2025-11-10 13:56:35 -08:00
• 40b26b456b Simplify the BatchMultimodalOutput in io_struct.py (#12993) Lianmin Zheng 2025-11-10 13:54:56 -08:00
• 9840bf4f84 [router][ci] Fix maturin build (#13012) Keyang Ru 2025-11-10 12:33:32 -08:00
• 303cc957e6 chore: bump SGLang version to 0.5.5.post1 (#13000) sglang-bot 2025-11-11 03:53:43 +08:00
• 661c1c97ad Add pre-suffle weight for new aiter MoE support. (#12908) 3 people 2025-11-11 03:42:55 +08:00
• c022107f8b Resolve HF download issue and download models before CI run starts for 8-gpu-h200 runners (#12952) Kangyan-Zhou 2025-11-10 11:32:05 -08:00
• 56c83e0fb3 [CI] Limit the CI trigger frequency of low-privilege actors (#13010) Liangsheng Yin 2025-11-11 03:27:10 +08:00
• b51d46d092 [AMD CI] Remove SRT docker build. (#11850) Sai EnduriandHubert Lu 2025-11-10 11:25:37 -08:00
• 1086473111 Enhance retract test (page cases, long output cases) (#12781) Liangsheng Yin 2025-11-11 03:03:26 +08:00
• 665416f6dd Unify memory management across (overlap, non-overlap) x (page>=1) x (spec, non-spec, spec v2) x (retract, finished) (#12224) Liangsheng Yin 2025-11-11 02:56:22 +08:00
• 838bcb0d93 [misc][ci] Add run-ci after auto-labeler (#13013) Chang Su 2025-11-10 10:41:54 -08:00
• f1f4c451ab Add process_prefill_chunk back to fix PP event loop (#13009) Liangsheng Yin 2025-11-11 00:51:36 +08:00
• b0ee99dd03 Super tiny fix typo (#13001) fzyzcjy 2025-11-11 00:47:45 +08:00
• ddfcb7c8ab minor: fix notebook bug with new model_info fields added for warmup (#13005) Mick 2025-11-11 00:46:12 +08:00
• 58b12ccb46 Support piecewise cuda graph for deepseek v3 (#12996) Ke Bao 2025-11-10 23:18:03 +08:00
• 547de8c774 [1 / 2] register weak_ref_tensor in sgl-kernel (#12999) Xiaoyu Zhang 2025-11-10 22:12:59 +08:00
• 37c40a87a8 chore: bump sgl-kernel version to 0.3.17 (#12966) sglang-bot 2025-11-10 21:50:58 +08:00
• 1240ac13b8 vlm: fix tiny multimodal cache bug (#12984) Yuhao Yang 2025-11-10 21:30:19 +08:00
• 5639145fac diffusion: reduce effort of supporting new model (#12982) Mick 2025-11-10 21:20:33 +08:00
• afee2843d5 feat(metrics): add scheduler and hiradix cache metrics (#10218) (#10225) ShawnKungandZhiqiang Xie 2025-11-10 18:14:47 +08:00
• 6f08488042 fix missing output_token_logprobs when using ngram speculative decoding (#10702) Zhihao Zhanganda4zhangfei 2025-11-10 18:04:42 +08:00
• 611a4fd08b [router] bucket policy (#11719) syy-hw 2025-11-10 18:02:53 +08:00
• 9ea2c686c7 [Auto Sync] Update batch_invariant_ops.py (20251109) (#12916) 3 people 2025-11-10 01:51:39 -08:00
• e2a784ecda [RadixTree] Reduce Syscalls, Optimize Collection Filtering and Align with cpp (#12239) PiteXChen 2025-11-10 17:47:59 +08:00
• 05559a4a90 Support hidden_dim % 4 == 0 in per_token_quant_fp8 (#12883) Xiaoyu Zhang 2025-11-10 17:13:14 +08:00
• 7bffc5dc25 [Fix] Add validation for served model name to reserve : for LoRA adapter syntax (#12912) Neelabh Sinhaandneelabhsinha 2025-11-09 23:42:02 -08:00
• a5e5088dfb Fix errors of page head kernels in sgl-kernel for ROCm (#12604) huangtingwei 2025-11-10 15:15:50 +08:00
• ac19ce7efb [PP] put pp assert in model runner (#12934) Xuchun Shang 2025-11-10 14:58:19 +08:00
• 95876d75cb chore: include a minimum image for vlms when warming-up (#9528) Mick 2025-11-10 14:56:59 +08:00
• a30f190762 chore: bump sgl-kernel version to 0.3.17 (#12931) sglang-bot 2025-11-10 14:54:02 +08:00
• dc8a5a1ce7 [Refactor / Style] Unify all event loops (except for PP) (#12959) Liangsheng Yin 2025-11-10 14:49:12 +08:00
• e123648b36 diffusion: fix wan-2.2-TI2V and support sp (#12926) Mick 2025-11-10 14:37:57 +08:00
• 90401cf7d2 Fix the run-time error when calling fused_rms_mxfp4_quant that change return output number (#12803) kkandwunhuang 2025-11-10 13:48:36 +08:00
• 9cfe78dd30 clean redundant code in previous PR (#12957) Atream 2025-11-10 13:40:32 +08:00
• 307e7a6128 diffusion: fix detected file changes rule in CI (#12943) Mick 2025-11-10 13:37:16 +08:00
• ddd1440d0f Refactor KTransformers heterogeneous compute with unified GPU-quantization backend (#12834) 5 people 2025-11-10 13:06:32 +08:00
• d1be60c3c5 [Refactor] rename set_index_k_and_scale_buffer to set_index_k_scale_b… (#12956) Edwin Gao 2025-11-09 23:35:19 -05:00
• 583bb1804e [Docs] Add docs for Qwen3-VL image and video support (#12554) Adarsh ShirawalmathandUbuntu 2025-11-10 09:46:04 +05:30
• 1f2a6c691b Bugfix: LMCache Connector with Sglang (#12946) MMuzzammil1andgemini-code-assist[bot] 2025-11-10 13:13:37 +09:00
• 61c7fe7aed Minor code cleanup / improvement for PREBUILT_EXTEND mode (#12948) Liangsheng Yin 2025-11-10 11:59:54 +08:00
• 83f89cc615 diffusion: skip full CI suite for multimodal_gen changes (#12940) Mick 2025-11-10 10:46:24 +08:00
• db24d34603 Support piecewise cuda graph for MLA (#11812) Ke Bao 2025-11-10 09:13:48 +08:00
• 885cfca273 ci: try to fix gpg error during kernel build (#12928) ishandhanani 2025-11-09 11:45:29 -08:00
• 4e916f9840 [lint] tiny fix unimported packages. (#12927) Liangsheng Yin 2025-11-10 00:59:10 +08:00
• 4f65a64666 Refactor / Unify event loop across PD-Disagg, Overlap, DP-Attn cases (#12839) Liangsheng Yinandcctry 2025-11-10 00:42:50 +08:00
• f5b3ccd9a5 feat: basic support for server-level multimodal cache (#10775) Mick 2025-11-10 00:27:50 +08:00
• bb00e24f87 Adjust server launch time in ci (#12917) Ke Bao 2025-11-09 20:41:29 +08:00
• 210a9cab6d [CI] Fix matrix.part in pr-test. (#12920) Liangsheng Yin 2025-11-09 18:53:18 +08:00
• b8ac4fcb51 [PD] feat: refactor custom mem pool and add barex pd support (#12332) Teng Ma 2025-11-09 18:51:34 +08:00
• 877cb52840 [CI] increase ut buckets & adjust estimation time. (#12919) Liangsheng Yin 2025-11-09 18:17:50 +08:00
• 3633f8b0cf Add Jet-Nemotron (#12448) Zijian Zhang 2025-11-09 17:32:47 +08:00
• 93cf60fc64 Fix Deepseek nightly tests (#12906) Kangyan-Zhou 2025-11-09 00:26:06 -08:00
• b5e0417392 Add kimi k2 thinking to ci (#12907) Ke Bao 2025-11-09 16:10:32 +08:00
• 8a821af793 fallback to triton mm_persistent kernel when deepGemm fail (#12911) Minglei Zhu 2025-11-08 23:42:26 -08:00
• 4b1d163bfd Add HF cleanup logic in ci_install_dependency.sh (#12895) Kangyan-Zhou 2025-11-08 22:56:25 -08:00
• c21a3ec299 Fix duplicate nightly test name (#12905) Kangyan-Zhou 2025-11-08 21:38:48 -08:00
• b142831a26 Fix empty server args in marlin moe test (#12904) Ke Bao 2025-11-09 13:30:47 +08:00
• d134096319 Add Deepseek models into nightly tests (#12865) Kangyan-Zhou 2025-11-08 21:17:49 -08:00
• f290e8016e Revert "Fix spec decoding acc length for dpsk-r1-fp4 tp8" (#12900) Qiaolin Yu 2025-11-08 20:42:42 -08:00
• 9299a62fcb Fix spec decoding acc length for dpsk-r1-fp4 tp8 (#12896) Qiaolin Yu 2025-11-08 19:55:45 -08:00
• 49543be9da Tiny simplify can_run_dp_cuda_graph gather logic (#12891) Liangsheng Yin 2025-11-09 11:15:36 +08:00
• 5236290399 Update CODEOWNERS (#12897) Ke Bao 2025-11-09 09:37:11 +08:00
• b2b26d4324 chore: bump sgl-kernel version to 0.3.16.post6 (#12889) sglang-bot 2025-11-09 09:16:28 +08:00
• d3a03aeef8 Refs/heads/add nightly test multi gpu configs (#12870) alisonshao 2025-11-08 15:14:50 -08:00
• 5f02b918ec [Fix] Fix trtllm-mla backend when chunked prefix cache is disabled (#12361) Baizhou Zhang 2025-11-08 15:10:25 -08:00
• 49653c8896 use fast stream instead of torch.cuda.current_stream in llama 4 shared experts overlap (#12811) b8zhong 2025-11-08 15:04:37 -08:00
• 44f594d832 Apply moe_reduce_sum kernel for fused_marlin_moe (#12888) Ke Bao 2025-11-09 01:31:05 +08:00
• 2b6c4257a0 Fix sending all requests to the first rank in DP attention (#12832) fzyzcjy 2025-11-09 00:18:53 +08:00
• 243ea585fc [DP-Attn] Clarify MLP sync / idle batch preparation logic (#12843) Liangsheng Yin 2025-11-08 23:23:14 +08:00
• 6fee2c535c [CI] Tiny adjust CI esitmation time (#12886) Liangsheng Yin 2025-11-08 23:02:19 +08:00
• f1a9c72de3 Support capturing aux_hidden_states for minimax m2. (#12798) Charles Chen 2025-11-08 01:54:22 -08:00
• 190002c613 [Docs][DeepseekV3.2] Update deepseekv3.2 docs for mha short seq prefill (#12868) YAMY 2025-11-08 00:11:02 -08:00
• 0296f1cdad [Auto Sync] Update activation.py, logits_processor.py, rota... (20251107) (#12853) 3 people 2025-11-07 22:07:51 -08:00
• e039ff382c [CI] Fix huggingface access for test_flash_attention_4.py (#12846) Baizhou Zhang 2025-11-07 20:07:06 -08:00
• b8ddc296f4 [sgl-kernel][Deepseek V3.2] Add row_starts to topk kernel (#12582) hlu1 2025-11-07 18:33:27 -08:00
• 0b88d520a0 Add nightly performance test for GPT-OSS 4GPU models (#12805) alisonshao 2025-11-07 16:54:07 -08:00
• d3d7f960b5 [router] Switch MCP tests from DeepWiki to self-hosted Brave search server (#12849) Keyang Ru 2025-11-07 16:16:07 -08:00
• e434187289 [router][grpc] Move all error logs to their call sites (#12859) Chang Su 2025-11-07 15:55:34 -08:00
• fe19a580fb [router][grpc] Refactor: Add builders for chat and responses (#12852) Chang Su 2025-11-07 15:43:32 -08:00
• 55e8e3999c add back flashinfer jit cache to dev docker (#12851) b8zhongandBrayden Zhong 2025-11-07 14:51:24 -08:00
• 32f7982800 sglang diffusion announcement (#12856) Mingyi 2025-11-07 14:35:25 -08:00
• 0f76976c3c remove the fa4 page_size hardcode to 128 restriction on mla model arch (#12801) Rain Jiang 2025-11-07 13:30:30 -08:00