Commit Graph

4193 Commits

Author SHA1 Message Date
Chang Su
d8ee15643b [Feat] Add reasoning parser for Qwen/Qwen3-235B-A22B-Thinking-2507 (#8363) 2025-07-25 14:59:42 -07:00
Yineng Zhang
7181ec8cfc fix: upgrade nccl version (#8359) 2025-07-25 14:59:02 -07:00
Lianmin Zheng
ed2e313eb6 Clean up server_args, triton cache manager (#8332) 2025-07-25 14:14:51 -07:00
Chang Su
f8260f2539 [Bugfix][Feat] Add XML-ish grammar in EBNFComposer and fix misc bugs in Qwen3 detector (#8357) 2025-07-25 12:03:16 -07:00
Xu Wenqing
12cb760a37 Add H20-3e fused MoE kernel tuning configs for Qwen3-Coder-480B-A35B-Instruct (#8344)
Signed-off-by: 许文卿 <xwq391974@alibaba-inc.com>
2025-07-25 10:58:12 -07:00
Stepan Kargaltsev
1b9cea5ade [P/D] Support ipv6 in P/D scenario (#7858)
Co-authored-by: Shangming Cai <caishangming@linux.alibaba.com>
2025-07-25 08:53:30 -07:00
Xiaoyu Zhang
9045cc1eb8 [torch.compile bug] avoid biased_grouped_topk_impl func repeatedly triggering torch.compile in forward pass (#8353) 2025-07-25 21:17:47 +08:00
Shangming Cai
70e37b97bf chore: upgrade mooncake 0.3.5 (#8341)
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
2025-07-25 01:17:26 -07:00
Zaili Wang
15d2759174 [CPU] Add tutorial docs for SGL on CPU (#8000) 2025-07-25 00:03:16 -07:00
Hubert Lu
af4b9bae95 [AMD] Add silu_and_mul, gelu_and_mul, gelu_tanh_and_mul, and gelu_quick kernels for AMD GPUs (#7135)
Co-authored-by: yiakwy-xpu-ml-framework-team <961186938@qq.com>
Co-authored-by: HAI <hixiao@gmail.com>
2025-07-24 23:44:28 -07:00
Ying Wang
7ad6b766c5 fix: Fix failed functional tests https://github.com/meta-llama/llama-stack-evals (#8266) 2025-07-24 23:11:32 -07:00
Cheng Wan
c0fb25e949 DP Enhancement (#8280) 2025-07-24 21:36:21 -07:00
li haoyang
28d4d47280 [Feature] Integrate quick allreduce and select the best allreduce implementation (#6619)
Signed-off-by: Haoyang Li <Haoyang.Li@amd.com>
Co-authored-by: ilmarkov <imarkov@redhat.com>
2025-07-24 20:48:42 -07:00
ZhichenJiang
f4674df646 support idle batch for TBO (#8233) 2025-07-24 20:43:52 -07:00
Zhiqiang Xie
d40846d456 breakdown kernel update (#8334) 2025-07-25 08:33:17 +08:00
Zhiqiang Xie
145482f422 HiCache Storage TP Refinement (#8307)
Co-authored-by: pansicheng <sicheng.pan.chn@gmail.com>
2025-07-25 08:31:47 +08:00
Simo Lin
39fe1e880d [router] add request format unit test (#8300) 2025-07-24 14:30:37 -07:00
Simo Lin
33c4b4d04e [router] add streaming unit test (#8299) 2025-07-24 14:30:27 -07:00
Swipe4057
8d1c5b948e chore: upgrade flashinfer v0.2.9rc1 (#8301)
Co-authored-by: Yineng Zhang <me@zhyncs.com>
2025-07-24 14:29:56 -07:00
Xiaoyu Zhang
a167fd0bcb [code style] Clean dead triton kernel code in fused_moe and useless vllm_ops import (#8310) 2025-07-24 14:38:30 +08:00
Simo Lin
2f86f3ad62 [router] add endpoint unit test (#8298) 2025-07-23 23:26:44 -07:00
Minho Ryu
bfb118c01e fix bug when eos_ids==0 (#8315) 2025-07-23 23:18:47 -07:00
Simo Lin
f6e07f2796 [router] fix pd model completion request (#8303) 2025-07-23 23:18:29 -07:00
Simo Lin
5dd0f870ab [bug] fix pd completion protocol for batching support (#8317) 2025-07-23 23:18:17 -07:00
Haohui Mai
f7e102d56a Pin the version of petit kernel to fix the APIs (#8235) 2025-07-23 17:57:20 -07:00
michael-amd
0e5fa67773 [AMD] Pull latest image for AMD CI (#8070) 2025-07-23 17:56:14 -07:00
xianzhiT
624a3b8d1f Fix incomplete tool call capture issue in streaming response of DeepSeek-V3 when enable MTP (#7562) 2025-07-23 17:40:23 -07:00
Chang Su
01079e174f feat(function call): complete utility method for KimiK2Detector and enhance documentation (#8043) 2025-07-23 17:37:31 -07:00
J
0e7a5b2694 fix: prevent crashes due to logit bias dimension mismatch (#7685) 2025-07-23 15:30:55 -07:00
Yineng Zhang
4953f4ca9a chore: upgrade sgl-kernel 0.2.7 (#8304) 2025-07-23 15:07:27 -07:00
Xinyuan Tong
38000a5f44 Fix gemma3n with hybrid swa (#8240)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2025-07-23 13:29:18 -07:00
Xinyuan Tong
70251e935e fix: match chat-template for internvl3 (#8262)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2025-07-23 13:29:03 -07:00
xianzhiT
c87d4fec99 Fix the issue of incorrect finish reason in final stream response chunk returned during tool call (#7708)
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
2025-07-23 13:28:53 -07:00
YiXR
a99801e075 [Performance][PD Disaggregation] optimize TokenToKVPoolAllocator by sorting free pages (#8133)
Signed-off-by: Xingrui Yi <yixingrui@linux.alibaba.com>
Co-authored-by: Xingrui Yi <yixingrui@linux.alibaba.com>
2025-07-23 13:28:12 -07:00
Yineng Zhang
4c605235aa fix: workaround for deepgemm warmup issue (#8302) 2025-07-23 12:01:51 -07:00
Simo Lin
6f8f4aeea4 [router] add common ut infra to mock worker and app (#8295) 2025-07-23 10:07:51 -07:00
Yuan Luo
0c8dab9e67 [sgl-kernel] Opt per_token_quant_fp8 with warp reduce (#8130)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2025-07-23 21:22:59 +08:00
Zhiqiang Xie
f39037fffb HiCache Fix (#8288)
Co-authored-by: pansicheng <sicheng.pan.chn@gmail.com>
2025-07-23 16:51:32 +08:00
Zhiqiang Xie
ce86e201df bug fix and tag (#8282) 2025-07-23 16:50:31 +08:00
Zhiqiang Xie
b43263307f Hicache IO kernel refactoring (#8264) 2025-07-23 16:49:03 +08:00
Lifu Huang
8abd3e77fe Introduce Stable LoRA ID System for Overlapped Updates and Prefix Caching (#8261) 2025-07-23 00:32:16 -07:00
Ke Bao
e885bfdc6a Fix sgl-kernel ci test (#8284) 2025-07-23 14:01:47 +08:00
Ke Bao
e2d66f60c8 Skip llama4 vision module loading when multimodal disabled (#8272)
Co-authored-by: Mick <mickjagger19@icloud.com>
2025-07-23 12:41:25 +08:00
Yineng Zhang
01c000043c chore: bump v0.4.9.post3 (#8265) 2025-07-22 15:55:48 -07:00
yhyang201
0dfe2491ac Preliminary Support for Qwen3XMLDetector (#8260)
Co-authored-by: Chayenne <zhaochen20@outlook.com>
2025-07-23 06:49:38 +08:00
zhongwei
ff45ab7a5f [Benchmark] add disable-auto-run param for hicache/bench_multiturn (#7822)
Co-authored-by: zhongwei.ren <zhongwei.ren@bytedance.com>
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
2025-07-22 14:02:40 -07:00
Peter Pan
0f8b538614 [fix] benchmark : routed_scaling_factor is None (#8059)
Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>
2025-07-22 08:55:35 -07:00
Rui Chen
c33499a67b fix: sgl-router remove dead code (#8257) 2025-07-22 08:41:23 -07:00
Hubert Lu
e50109f2ed [AMD] Remove vllm's scaled_fp8_quant and moe_sum when SGLANG_USE_AITER=1 (#7484) 2025-07-21 17:33:19 -07:00
Xinyuan Tong
69adc4f81c fix: retrieve mm token by modality, raise error if none (#8221)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2025-07-22 08:06:35 +08:00