chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Commit Graph

  • 547e2d037e [diffusion] refactor: add arg to control the precision of dit (#17751) fsygd 2026-01-27 23:01:23 +08:00
  • 1d942e4eef [DeepSeek] Update tests and document for DeepSeek V3.2 NVFP4 checkpoint (#17657) Baizhou Zhang 2026-01-27 22:10:57 +08:00
  • d578b41bad [NPU] Adapt cann 8.5: use sfa and lightning indexer op from cann and CI update (#17615) monkeyLoveding 2026-01-27 19:03:53 +08:00
  • 832c756549 [Doc] Tiny update description on torch compile (#17819) Baizhou Zhang 2026-01-27 18:59:04 +08:00
  • c56d19b977 fix(quantization): add sgl_kernel fallback for FP4 quantize on Blackwell GPUs (#17816) MikkoParkkola 2026-01-27 11:43:17 +01:00
  • dba264ac73 [PP] fix wrong weight logic for tie_word_embeddings model (#15890) Xuchun Shang 2026-01-27 17:41:17 +08:00
  • 7106f6c8e1 [GLM-OCR] Support GLM-OCR Model (#17582) Yuxuan Zhang 2026-01-27 14:24:00 +08:00
  • 81c0f5c5ad [Model] Add support for EXAONE-4.0 Model (#8205) Taemin Jung 2026-01-27 15:08:24 +09:00
  • 6c9b054ab7 [Bug Fix] Fix reasoning parser when continue_final_message=true (#17065) laixin 2026-01-27 14:04:44 +08:00
  • df42f4d386 [AMD] Update dsv3.2 AMD GPU docs and unify ROCm TileLang build (#17783) Hubert Lu 2026-01-26 21:10:32 -08:00
  • a723d1c5ef [model-gateway] ignore error for embeddings/classify in PD router (#15931) shuwenn 2026-01-27 11:49:14 +08:00
  • 57e432d951 fix: preserve disconnect events in api key middleware (#17253) shuwenn 2026-01-27 11:48:24 +08:00
  • fd3b179ffd [HiCache][HA 1/N] Support HiCache storage runtime attach/detach (#15892) shuwenn 2026-01-27 11:33:19 +08:00
  • 1b56a886bb [chore]: improve time tracing of model loading process (#15426) Zhongdongming Dai 2026-01-26 19:04:25 -08:00
  • 3ad3268e06 [CI] Skip PD hybrid attention test with different TP temporarily (#17791) Shangming Cai 2026-01-27 10:59:33 +08:00
  • 479ab7a4e7 model: support Kimi-K2.5 (#17789) Yuhao Yang 2026-01-27 10:57:00 +08:00
  • 0519b0935f [diffusion] comfyui: support Qwen-Image, Multi-GPU Z-Image, and Enhanced ComfyUI Integration (#17678) WenhaoZhang 2026-01-27 10:06:42 +08:00
  • 2d8c22a15e [bugfix] Internal processing of hf3fs crash # 16614 (#16938) FlyPanda 2026-01-27 10:01:50 +08:00
  • 539924037f fix(processor): support InternS1 text_config in InternVL processor (#17040) Mahdi-CV 2026-01-26 13:02:54 -08:00
  • dd97e1fe38 [Docs] Add RL documentation (#17663) zijiexia 2026-01-26 12:16:54 -08:00
  • 5ab76ff220 Special logic for healthcheck (#17734) ybyang 2026-01-27 02:26:40 +08:00
  • 85d077f44d Introduce global alloc_len_per_decode & clean check decode memory (#15115) Liangsheng Yin 2026-01-26 10:26:20 -08:00
  • 8643fb2f52 fix: remove truncation for test and job names in ci failure monitor (#17765) Douglas Yang 2026-01-26 09:46:33 -08:00
  • bba6e38ff8 [NPU] Split pyproject npu from pyproject other (#17641) Makcum888e 2026-01-26 20:45:44 +03:00
  • 51d139b867 fix: move nightly whl to cuda version folder (#17762) Douglas Yang 2026-01-26 08:13:46 -08:00
  • 7bb41989fa [1/N] Optimize All Reduce - Benchmark different AR operations (#13797) Yuan Luo 2026-01-26 22:44:13 +08:00
  • 6c0f9b4824 Add test_gpt_oss_4gpu.py to B200 test suite (#17743) Alison Shao 2026-01-26 05:57:06 -08:00
  • b56366f827 [NPU]DeepSeek-V3.2 support npu mlaprolog (#15381) lawtherWu 2026-01-26 20:42:37 +08:00
  • 738b1ac988 [AMD CI] Add moonshotai/Kimi-K2-Instruct-0905 testcases (#17656) sogalin 2026-01-26 18:12:34 +08:00
  • 5844cb2fd8 refactor mamba radix cache logic in server_args (#17645) Yi Zhang 2026-01-26 17:02:49 +08:00
  • f6f1b6d000 Bump FI version (#17700) shaharmor98 2026-01-26 10:50:06 +02:00
  • 2734b23481 accuracy enhancement for baichuan2-13B for npu (#16868) McZyWu 2026-01-26 16:14:35 +08:00
  • 6756cf18c6 remove multimodal as this is completely dead code (#17750) Simo Lin 2026-01-26 03:07:40 -05:00
  • 12f794e516 [diffusion] fix: fix missing backend argument in pipelines_core initialization (#17343) Prozac614 2026-01-26 15:47:10 +08:00
  • d4adff31aa update wasm endpoint (#17748) Simo Lin 2026-01-26 02:43:29 -05:00
  • 48f4340b14 Exclude some diffusion package for ARM in docker release (#17745) Kangyan-Zhou 2026-01-25 23:32:39 -08:00
  • ed75136e85 remove self managed wasm as it has been replaced with official smg wa… (#17746) Simo Lin 2026-01-26 01:56:33 -05:00
  • 30b3192039 Merge performance/accuracy test suites into regular stage-b suites (#17609) Alison Shao 2026-01-25 22:49:19 -08:00
  • 8c2d8b51e9 [model-gateway] fix wasm example2 (#17244) Praneth Paruchuri 2026-01-26 12:15:36 +05:30
  • 02c1dabf5d [model-gateway] fix wasm example3 (#17277) Praneth Paruchuri 2026-01-26 12:15:21 +05:30
  • 511961870f [model-gateway] Optimize WASM cache lookups using SHA-256 (#17344) Praneth Paruchuri 2026-01-26 12:14:55 +05:30
  • 1a19b3987d [Model] Add Ernie4.5 VL model support (#15679) CSWYF3634076 2026-01-26 14:36:29 +08:00
  • d275d47973 [NIXL] Add custom NIXL backend selection for KVManager (#17146) zackyoray 2026-01-26 08:35:38 +02:00
  • 1e8db18290 [Kimi-Linear] Remove duplicated code in kimi-linear (#17731) Yuan Luo 2026-01-26 14:20:24 +08:00
  • 7890a96f97 remove self managed mcp as it has been replaced with official rmcp crate (#17740) Simo Lin 2026-01-26 01:17:06 -05:00
  • 52e0f65fce Update nightly-test-nvidia.yml to remove push trigger (#17625) Kangyan-Zhou 2026-01-25 21:42:36 -08:00
  • 5aaedac3c6 Add EP=2 to qwen235b nightly tests (#17738) Kangyan-Zhou 2026-01-25 21:36:04 -08:00
  • 7b22b8ff8a Fix sgl-kernel install: fail instead of PyPI fallback when artifacts missing (#17728) Alison Shao 2026-01-25 19:46:49 -08:00
  • 444b9521e4 [Bugfix]Repeated add modelslim quant_config and bugfix with "enable-piecewise-cuda-graph" on NPU (#17511) chenxu214 2026-01-26 09:51:07 +08:00
  • 46bc53a81b [smg] import db crate to replace self managed one (#17727) Simo Lin 2026-01-25 19:54:52 -05:00
  • 52d0ca944b [smg] import official wfaas crate to replace self managed one (#17724) Simo Lin 2026-01-25 18:20:58 -05:00
  • 592603d77b Fix flaky streaming logprobs test by handling detokenizer text buffering (#17687) Kangyan-Zhou 2026-01-25 15:09:06 -08:00
  • bf139af37a [smg] use already published auth crate for better compilation speed (#17723) Simo Lin 2026-01-25 17:52:42 -05:00
  • 344eeaee90 Upload nightly test metrics to GH artifacts (#17696) Kangyan-Zhou 2026-01-25 14:35:14 -08:00
  • 97a36a72b7 [smg] remove dead tokenizer code (#17722) Simo Lin 2026-01-25 17:04:48 -05:00
  • fc7096f80b [model-gateway] Optimize special token search using Aho-Corasick (#17387) Praneth Paruchuri 2026-01-26 03:24:15 +05:30
  • f5ac1ca10b [smg] use official tokenizer crate instead of manually built one (#17721) Simo Lin 2026-01-25 16:35:16 -05:00
  • 2c1e2674cc [misc] replace existing tool call code with new crate package (#17720) Simo Lin 2026-01-25 16:07:39 -05:00
  • 48f5c46a8d [misc] remove tool parser and tree benchmark as they are not meaningful atm (#17719) Simo Lin 2026-01-25 16:03:06 -05:00
  • 7ca8c12e0e Extend b200 kernel tests timeout for CPU differences (#17718) Kangyan-Zhou 2026-01-25 12:59:39 -08:00
  • 86c7bc6431 [smg] update crate for tools (#17717) Simo Lin 2026-01-25 15:56:42 -05:00
  • 8d3e1ac0c8 Add an all type in pyproject.tml to include diffusion support (#17697) Kangyan-Zhou 2026-01-25 12:52:13 -08:00
  • 9123491430 A few updates to the night tests (#17694) Kangyan-Zhou 2026-01-25 11:20:17 -08:00
  • a883906a24 Support mxint4 flashinfer_trtllm moe gemm (#16892) HandH1998 2026-01-26 00:15:53 +08:00
  • b105dad5da [diffusion] refactor: remove useless lazy-import cache-dit codes (#17659) Mick 2026-01-25 22:43:22 +08:00
  • 7a1d7ab47b remove self managed protocols as it has been replaced with official oai spec (#17711) Simo Lin 2026-01-25 09:38:13 -05:00
  • 8db2802b2d update to use official openai protocol crate (#17710) Simo Lin 2026-01-25 09:35:59 -05:00
  • 6af22f8dbf use published reasoning parser crate (#17709) Simo Lin 2026-01-25 09:15:03 -05:00
  • fb61164f27 [Refactor] Use is_in_ci() utility in JIT kernel benchmarks (#17118) Zhengbo Wang 2026-01-25 20:40:47 +08:00
  • 64d809937a revert row from https://github.com/sgl-project/sglang/pull/17584/ (#17701) Makcum888e 2026-01-25 12:17:47 +03:00
  • d1042e0d62 [Refactore] [CI] Remove redundant CI test runs step 2 (#17584) Makcum888e 2026-01-25 10:39:48 +03:00
  • 9bd92ba0f6 Support PD disaggregation with different TP/DP size for Qwen3-Next (#16056) xjx471258437 2026-01-25 15:34:02 +08:00
  • 30ece5e1d6 Fix swa memory pool size with spec (#17630) Ke Bao 2026-01-25 14:10:43 +08:00
  • 1674b9ef44 [DeepSeek-V3.2] Fix TRT-LLM NSA in target_verify/draft_extend (#17662) Mohammad Miadh Angkad 2026-01-25 13:10:14 +08:00
  • 9121f22656 Add PyTorch .bin file validation to CI weight validation (#17533) Alison Shao 2026-01-24 19:18:15 -08:00
  • b829b797ef Fix slash command handler trigger condition by trimming the comments (#17691) Kangyan-Zhou 2026-01-24 18:54:46 -08:00
  • 59f027a8c8 [diffusion]: Fix ZImage SP sharding for caption and latent (#17301) Chen Shen 2026-01-25 10:10:48 +08:00
  • 37c04c2245 fix: Refactor register_image_processor to use kwarg instead of positional arg (#17685) Xinyuan Tong 2026-01-24 18:31:01 -05:00
  • 69a7a70e47 Temporarily disable lora overlap loading test due to flakiness (#17683) Kangyan-Zhou 2026-01-24 12:24:44 -08:00
  • 137eb5b95c Fix NSA indexer test and move it to pre commit test (#17682) Kangyan-Zhou 2026-01-24 12:06:18 -08:00
  • 8656a146a6 Fix test timeout issue in pr-test (#17681) Kangyan-Zhou 2026-01-24 11:46:08 -08:00
  • 2c2c4e446b [NVIDIA] Add flashinfer all-to-all MOE dispatcher (#14668) Trevor Morris 2026-01-24 06:59:55 -08:00
  • 458a43d4ac [NPU] torch_npu profiler tensorboard path type fix (#17545) TMC 2026-01-24 22:55:49 +08:00
  • 0c8165ffbd [Kimi-Linear] Refactor Kimi-Linear to support RadixLinearAttention (#17506) Yuan Luo 2026-01-24 21:27:13 +08:00
  • bf19d20d89 [Bugfix] fix TypeError when log-requests-level >=2 in prefill node warmup (#17129) yunkchen 2026-01-24 19:16:22 +08:00
  • 0834f9afeb [Auto Sync] Update test_deterministic.py (20260124) (#17665) Lianmin Zheng 2026-01-24 02:52:47 -08:00
  • a6280b2a23 add documentation example for LoRA overlap loading and cleanup unused function (#17464) Glen Liu 2026-01-24 02:33:16 -05:00
  • 3992a023e6 Move fa4 from sgl-kernel to jit kernel (#17353) Xiaoyu Zhang 2026-01-24 15:25:03 +08:00
  • 7a4bb0d516 [Diffusion] Add diffusion time embedding to jit kernel (#17658) Xiaoyu Zhang 2026-01-24 14:27:08 +08:00
  • fb683be6eb Use attn tp group in embedding for more models (#17570) Ke Bao 2026-01-24 13:37:44 +08:00
  • 176da1bbdd Fix: mistake sigmoid in kda (#17508) strgrb 2026-01-24 13:35:14 +08:00
  • 4c512a7d1d [JIT Kernel]Add Some CUDA Runtime API Wrapper for JIT Kernel Header (#17588) Qi Yuhang 2026-01-24 12:57:58 +08:00
  • 58799d91a7 fix: nightly wheel naming for non-post versions (#17538) Douglas Yang 2026-01-23 20:51:52 -08:00
  • d0919be733 [diffusion] model: LTX-2 Support (2/2) (#17496) GMI Xiao Jin 2026-01-23 20:51:37 -08:00
  • 797a9811a2 [diffusion] model: LTX-2 (1/2) (#17495) GMI Xiao Jin 2026-01-23 19:59:48 -08:00
  • 894928a951 Refactor: Extract DeepSeek common utilities into shared module (#16969) Ananya 2026-01-24 08:59:52 +05:30
  • 0dfe46dafb [Docker] Install cudnn==9.16 for cuda 13 image to avoid check error (#17668) Baizhou Zhang 2026-01-24 11:27:03 +08:00
  • bc6f0b5ce7 [Auto Sync] Update logits_processor.py, test_logprobs.py (20260124) (#17664) Lianmin Zheng 2026-01-23 17:57:41 -08:00
  • e1833c4f5a Add yeahdongcn to CI permissions (#17667) R0CKSTAR 2026-01-24 09:56:09 +08:00
  • b4a611fb33 [NPU] solve accuracy problem for stablelm-2-1-6b for npu (#17470) McZyWu 2026-01-24 08:27:38 +08:00