Commit Graph

7855 Commits

Author SHA1 Message Date
kousakawang
7efd8b3d1f [FEAT] Shared mem pool based cuda ipc for multi-modal data transport (#11917)
Co-authored-by: kousakawang <wanghanpei@bytedance.com>
Co-authored-by: Yuan Luo <4908075+yuan-luo@users.noreply.github.com>
2025-11-02 16:46:37 +08:00
Ho-Ren (Jack) Chuang
76196b3cbf feat: Add FP4 (E2M1) KV Cache Support with Quantization Utilities for MLA (#10078)
Signed-off-by: Ho-Ren (Jack) Chuang <horenchuang@bytedance.com>
Co-authored-by: Yichen Wang <yichen.wang@bytedance.com>
2025-11-01 22:24:58 -07:00
Binyao Jiang
3451fc3280 [Feature] Qwen3-Next & FLA: Support MTP topk>1; Up to 6% faster (#11133)
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
2025-11-01 19:47:56 -07:00
Zhihao Lyu
c550ab9125 [Ascend] Add Ascend NPU support for sglang.check_env & rework proposal (#11052)
Co-authored-by: ronnie_zheng <zl19940307@163.com>
2025-11-01 19:26:45 -07:00
Xun Sun
0afd68321b Update Mooncake EP's a2a interface (#12391) 2025-11-01 18:48:47 -07:00
Johnsonms
6f858930c8 [Bug] test_flashattn_mla_backend errors in Hopper #12487 (#12488) 2025-11-01 18:28:06 -07:00
hzh0425
6b634493c3 [HICache / PD]: Support offloading incremental KV cache in decode side. (#11966) 2025-11-01 14:59:37 -07:00
Xinyuan Tong
d2a8f71c2f [feat] Add SGLANG_TOOL_STRICT_LEVEL for tool-call behavior control (#12423)
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
2025-11-01 13:15:02 -07:00
Ke Bao
69193f7122 Filter tokenizer warning for kimi models (#12485) 2025-11-01 16:27:31 +08:00
yinghui
d5b6e50fe8 perf: trtllm mla performance minor improvements (#12435) 2025-10-31 22:48:02 -07:00
Liangsheng Yin
9632e48f5d [hot fix] Remove from python.sglang.xxx (#12483) 2025-11-01 11:00:05 +08:00
Qiaolin Yu
59cce5941a Use sgl fp4 quant kernel by default (#12482) 2025-10-31 19:51:28 -07:00
Surya-Gunukula
795e98f8a6 Forward unknown tool calls instead of dropping (#12226) 2025-11-01 02:10:35 +00:00
Shangming Cai
358ae3563d Tiny fix eos handling for PD disaggregation (#12334)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2025-10-31 17:57:10 -07:00
sglang-bot
41c10e67fc chore: bump SGLang version to 0.5.4.post2 (#12439) 2025-10-31 17:38:50 -07:00
Xinyuan Tong
0bfe1d145c fa3 & trtllm_mha spec overlap (#11874) 2025-10-31 17:38:13 -07:00
Ke Bao
a4bf5c6ad2 Support Kimi Linear (#12469)
Co-authored-by: yizhang2077 <1109276519@qq.com>
2025-10-31 14:03:35 -07:00
fzyzcjy
30ad107028 Try to allow NCCL cumem for multi node nvlink case (#11987) 2025-10-31 12:48:25 -07:00
Ke Bao
f7f9e41b36 Fix run benchmark (#12473) 2025-11-01 02:39:48 +08:00
ishandhanani
263eab9f5d fix: dummy health check server not accessible on non-zero rank nodes (#12297) 2025-10-31 11:34:57 -07:00
fzyzcjy
25257d8e00 Tiny assert no running requests when releasing memory to avoid IMA (#12341) 2025-11-01 01:28:53 +08:00
daniel, chen
cf0c24150a add served model name in bench serving (#12428) 2025-11-01 01:28:11 +08:00
huangtingwei
5538e05cb1 fix default env var for mooncake store (#12429) 2025-11-01 01:25:33 +08:00
Yuan Luo
c30ebb9300 [VLM] Optimize async mm data process mechanism (#12066)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2025-11-01 01:24:53 +08:00
ykcombat
41efcaeb45 [Feature] PD-Multiplexing Context and Scheduler, lazy import spatial. (#12275) 2025-11-01 00:40:01 +08:00
0xNullPath
70562969b9 [Bug] OOM (Out-of-Memory) errors for extreme testing scenarios (min_tokens=2) (#11757)
Signed-off-by: Yan Lu <luyan@nvidia.com>
2025-11-01 00:28:41 +08:00
Ke Bao
0095e01874 Fix lint in deepseek-ocr (#12470) 2025-11-01 00:08:19 +08:00
Xinyuan Tong
684864814b Feat: deepseek-ocr logits processor (#12415)
Co-authored-by: xinyuant <xinyuant@usc.edu>
2025-10-31 23:35:22 +08:00
sjtu_shenhai
410225b719 [Bug fix] Fix severe memory waste issue with torch.empty pin_memory (#12266) 2025-10-31 21:30:37 +08:00
Liangsheng Yin
2c9aebea70 Simplify watchdog (#12463) 2025-10-31 21:17:38 +08:00
Kindyaa
bc741073a3 fix:watchdog thread exception (#12328) 2025-10-31 20:54:50 +08:00
Yuhong Guo
2f6af1a3de Enable bailing_moe to support TP=16 (#12369) 2025-10-31 19:32:49 +08:00
Zaili Wang
50b6842b4b fix: Add default value for backend in sample_mmmu_requests (#12256) 2025-10-31 19:31:40 +08:00
Liangsheng Yin
300b4c212e [CI] fix possible port conflicts. (#12452)
Co-authored-by: Lianmin Zheng <15100009+merrymercy@users.noreply.github.com>
2025-10-31 17:20:20 +08:00
Lianmin Zheng
c0652d907b Clean up sgl kernel (#12413)
Co-authored-by: Byron Hsu <byronhsu1230@gmail.com>
2025-10-31 01:13:34 -07:00
Bruce-x-1997
2e48584b62 [bugfix] set is_prefill_only=false when mixed_chunk (#10889) 2025-10-31 00:12:21 -07:00
Liangsheng Yin
5cc0d25a0d Add trait for BasePrefixCache (#12436) 2025-10-31 14:03:27 +08:00
b8zhong
a076ec1a7a Revert "fix llama4 kv cache layout" (#12437) 2025-10-30 22:33:37 -07:00
bppps
2f766f381d [Bugfix]: distinguish processors for deepseek_vl2 and deepseek_ocr to p… (#12384) 2025-10-30 21:51:30 -07:00
elvischenv
069e490bcb feat: support trtllm_mha FP8 query attention kernel (#12307) 2025-10-30 21:39:35 -07:00
Yuhong Guo
ab95d35fcb feat: Add Non-intrusive Tensor Dumping for Model Inference (#10566) 2025-10-31 12:04:48 +08:00
b8zhong
34c286b811 Fix the shared expert & routed expert overlap in Llama 4 (#12405)
Co-authored-by: Brayden Zhong <b8zhong@users.noreply.github.com>
2025-10-30 20:14:49 -07:00
Liangsheng Yin
9416ee60dd Reserved abortion API when retracting (#12425) 2025-10-31 10:56:05 +08:00
Johnsonms
662725b936 [DeepSeekV32] Bug fix to ensure page_table and result in same type (#12300) 2025-10-30 18:56:51 -07:00
lpc0220
6c1a3f0cb2 enable cudaProfilerApi for one batch benchmarking (#11116) 2025-10-30 15:43:59 -07:00
popsiclexu
7d121448b2 [Bug fix][PD Dissaggregation] fix prefill hanging issue with PP and DP Attention, (#12368) 2025-10-30 14:22:47 -07:00
Lianmin Zheng
6a63a9852e minor code sync (#12403) 2025-10-30 12:49:31 -07:00
Xiaoyu Zhang
04e5b6faa7 Revert "Triton fused_moe_kernel support ep moe tuning" (#12377) 2025-10-30 07:12:06 -07:00
Even Zhou
ce6b17c0f9 [Feature] Support DeepSeek MTP on NPU (#11897)
Co-authored-by: liupeng374 <liupeng374@huawei.com>
2025-10-30 21:56:56 +08:00
Even Zhou
cafebef154 [NPU] bugfix for Qwen3-Next and performance update (#11969) 2025-10-30 21:52:16 +08:00