Commit Graph

Select branches

Hide Pull Requests

bot/update-est-time-20260928

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• db84a8ebbb [Model] Qwen3ForPooledOutput: forward get_input_embeddings to inner model (#23434) jsheng_Linkedin 2026-04-29 12:25:06 -07:00
• 3272af2f00 [Apple Silicon] [MLX] MLX decode partial overlap scheduling for generation (async eval) (#22416) 3 people 2026-04-29 15:21:14 -04:00
• d4040e7010 [CI] Broaden stage-b-test-1-gpu-large runner pool to H100 + H200 (#24080) Kangyan-ZhouandClaude Opus 4.7 2026-04-29 12:18:10 -07:00
• 903e46d848 [Bench] fix bench_hf.py KeyError + reduce print spam + add --limit (#24079) Yihao Wang 2026-04-29 14:43:37 -04:00
• 1376761841 fix(moe): repair dead import in fused_moe_native after MoE refactor (#24069) Xinyuan Tong 2026-04-29 19:14:52 +01:00
• 530b497a48 [BugFix] correct host leaf status check from evicted to backuped (#23537) Jeongho Shin 2026-04-30 02:26:00 +09:00
• 03147f66b8 ci: add /rerun-group to rerun all registered tests in a group (#24023) shuwennandClaude Opus 4.7 2026-04-30 01:24:16 +08:00
• 4cf109bbd1 debug followup (#24058) Xinyuan Tong 2026-04-29 16:03:27 +01:00
• 3f7c95d6cc [JIT Kernel][1/2]Migrate MXFP8 Group GEMM & Quant into JIT (#23833) Qi Yuhangandgemini-code-assist[bot] 2026-04-29 22:50:09 +08:00
• eeb7b5c433 [diffusion] fix: align encoder of flux klein with official (#24008) Mick 2026-04-29 22:42:16 +08:00
• 155e333039 [AMD] Update AMD CI workflow concurrency group (#24065) YC Yen-Ching Tsengandbingxche 2026-04-29 22:39:18 +08:00
• e5c2a9b6cf [diffusion] fix: improve LTX2.3 reference accuracy controls (#24022) Mick 2026-04-29 21:39:27 +08:00
• 1279ae0787 Bugfix (#24027) Xinyuan Tong 2026-04-29 14:13:51 +01:00
• 4c1eefca4f [NPU] ascend backend support qwen3 moe attention cp (#21685) AndyLi429 2026-04-29 19:25:17 +08:00
• ae0c036c24 [BugFix][EPD] fix embedding req_id transfer error (#23481) Zheng Wengang 2026-04-29 18:56:03 +08:00
• 180bb2624f [AMD] Fix CI RuntimeError: opentelemetry package is not installed (#23940) jacky.chengandBingxu Chen 2026-04-29 18:02:44 +08:00
• d9270b8c6a fix(moe): relocate orphan tuned configs after #23019 (#24004) hhwxw 2026-04-29 05:00:13 -04:00
• 6c7b242181 mimo v2.5 pro sglang-jax cookbook (#23936) Brian 2026-04-29 16:58:41 +08:00
• 73e93bebd6 [1/4] NVFP4 KV cache: quantization strategy abstraction and kernel (#21954) Sam (Kesen Li) 2026-04-29 16:45:48 +08:00
• 8327270c72 [Kernel] Support FlashInfer TRTLLM-Gen fused MoE for non-gated FP4 & FP8 (Nemotron) (#21321) danielafrimi 2026-04-29 11:28:21 +03:00
• bd448e51bd [Spec] Split accept_length into num_accepted_drafts and num_accepted_tokens (#23962) Liangsheng Yin 2026-04-29 00:02:22 -07:00
• 832b4f59ed [Bench] fix MMMU answer-extraction regex dropping multi-line responses (#23864) Xinyuan Tong 2026-04-29 07:48:49 +01:00
• 2c41ef4c93 [HiCache] feat: add draft KV cache backing for L2/L3 (#21125) shuwenn 2026-04-29 14:47:31 +08:00
• 71d2227a78 [diffusion] CI: update ground truth with official output (#23714) Mick 2026-04-29 13:51:31 +08:00
• 5a94088178 Fix typo: page_first_kv_spilt -> page_first_kv_split (#23983) JieTang66andJieTang 2026-04-29 12:08:05 +08:00
• 23cfa9e4c8 [NPU] fix rope_theta get error for baichuan2-13b-chat model (#21543) ZeyuanChen2000 2026-04-29 12:01:03 +08:00
• b437f6be48 model: Nemotron-omni-v3-alias (#23857) Yuhao Yang 2026-04-29 11:08:23 +08:00
• f57ec8d6ef [spec decoding] add extra attribute 'spec_hidden_size' (#23890) Qiaolin Yu 2026-04-29 10:54:50 +08:00
• 2a771a40ac Add engine_type label to tokenizer manager metrics (#23978) Lianmin Zheng 2026-04-28 19:52:58 -07:00
• d66eb3a91b docs: update contribution guide with coding style guidelines (#23977) Lianmin Zheng 2026-04-28 19:51:58 -07:00
• 1aee04e7df [XPU] Support apply_router_weight_on_input for Llama4 for fused_experts (#22654) Rahul Vijayaraghavan 2026-04-29 08:14:48 +05:30
• 1e14bd6f36 Fix for CVE-2026-5760 (#23660) ccullen-cert 2026-04-28 22:39:44 -04:00
• 4e885baa9b docs(cookbook): add H200 (FP4) deployment option for DeepSeek-V4 (#23980) Baizhou Zhang 2026-04-28 19:38:53 -07:00
• feec1ac7f9 ci: clean up stale-CUDA mooncake variant in install_extra_deps (#23960) Kangyan-ZhouandClaude Opus 4.7 2026-04-28 19:32:38 -07:00
• e5da200d0a [AMD] Fix Aiter RMSNorm layout handling (#23974) Hubert Lu 2026-04-28 19:28:46 -07:00
• 0ac23cffac Add intel_xpu as backend for GptOssForCausalLM, enabled for bf16 models with torch native backend (#12771) Chandrakant KhandelwalandMa Mingfei 2026-04-29 07:51:52 +05:30
• 08699bb1b2 [NPU] Fix DeepEP LL dispatch BF16 flag and skip triton kernel on NPU for Qwen3.5 (#23815) iridiumineandiridiumine 2026-04-29 10:19:24 +08:00
• 2d27c38f13 Update XPU Docker runtime stack & hf_home config (#23820) MingxuZhandMa Mingfei 2026-04-28 19:03:11 -07:00
• 13afe8acdf [codex] Enable Qwen3-Next MoE all-reduce fusion (#23619) Xiaoyu Zhang 2026-04-29 09:11:35 +08:00
• 14b4e6fa69 Support --model as alias for --model-path in CLI (#23894) Lianmin Zheng 2026-04-28 17:23:27 -07:00
• 233048212a [HiCache][SPEC] fix: normalize storage prefetch key (#23631) shuwenn 2026-04-29 06:53:20 +08:00
• 387c932dfc [Docs] update Docker image for Nemotron 3 Nano Omni (#23968) zijiexia 2026-04-28 15:08:34 -07:00
• ddcacaf1bd Fix failing test_nvidia_nemotron_3_nano by fixing test_grouped_topk (#23874) Khoa Pham 2026-04-28 15:03:58 -07:00
• 345fecc547 fix(bench): wire request_func in bench_long_context ContextWorkloadGenerator (#23898) Alex NailsandClaude Opus 4.7 2026-04-28 14:45:51 -07:00
• cf0061da43 [Spec] Fix spec_accept_rate and unify accept/draft naming (#23530) Liangsheng Yin 2026-04-28 14:40:04 -07:00
• 3e1c5e1b74 [HiCache][SPEC] fix: empty key after page alignment in match_prefix (#23387) shuwennandClaude Opus 4.6 2026-04-29 05:06:55 +08:00
• 9814cc89ce [Fix] NVFP4 qwen3.5 quant error fix by add packed_modules_mapping (#23471) Lewisand百麒 2026-04-29 04:36:09 +08:00
• 914ef7c7f3 Fix multimodal /v1/embeddings Jinja chat template handling (#20835) 3 people 2026-04-29 04:05:45 +08:00
• dc1eac4903 [MUSA][Diffusion] Fix fa3 API on MT MUSA (#23646) Qingfu Wen 2026-04-29 04:01:35 +08:00
• 826f2d0620 chore(codeowners): add @kpham-sgl as owner for gemma4 files (#23916) Khoa PhamandClaude Opus 4.7 2026-04-28 11:43:32 -07:00
• 66ea0aee7f tokenizer: Add fastokens support (#23753) AlonKejzman 2026-04-28 21:43:10 +03:00
• ad785a2299 [Docs] add Nemotron 3 Nano Omni cookbook (#23907) zijiexia 2026-04-28 10:24:40 -07:00
• e458a9248f docs: enable MiMo V2.5 MTP cookbook path (#23945) Xinyuan Tong 2026-04-28 18:22:19 +01:00
• 3fce8f2009 [Docs] add cookbook for Ling-2.6 family (#23947) Xinyuan Tong 2026-04-28 17:42:04 +01:00
• d95715ec65 [AMD] Fix CI test_diffusion_generation[flux_2_image_t2i_2_gpus] (#23944) jacky.cheng 2026-04-28 23:06:29 +08:00
• 4e1ef6b3cf [Docs] Add single-node H200 DeepSeek-V4-Pro low-latency recipe (#23943) Yuhao Yang 2026-04-28 23:03:26 +08:00
• 144038fbae [diffusion] chore: change default seed to 42 (#23836) Mick 2026-04-28 20:39:23 +08:00
• 69a71219cb feat: tiny improve fp8_gemm tune usage (#23912) Muqi Li 2026-04-28 19:47:46 +08:00
• 7824903417 [SKILL] Sync SGLang skill docs (#23921) Xiaoyu Zhang 2026-04-28 17:05:36 +08:00
• 71160e4ddb feat(observability): add OpenTelemetry tracing for pipeline parallelism (#23169) Yinzuo Jiang 2026-04-28 17:05:23 +08:00
• 9a53ab3d6d [6/N] (Elastic EP) Recover failed ranks (#15771) Xun Sun 2026-04-28 15:44:26 +08:00
• b8a2dcd300 fix: resolve tensor file overwrite between target and draft models (#21694) 3 people 2026-04-28 14:40:49 +08:00
• 7ce2352323 [docs] fix sglang serve --model-path in cookbooks (#23905) zijiexia 2026-04-27 21:54:21 -07:00
• 6fbad22feb Remove smoke wording from tests and comments (#23355) Xiaoyu Zhang 2026-04-28 12:05:27 +08:00
• 1a55646dcd [Feature] Xiaomi MiMo-V2.5-Pro day0 support (#23808) JoyFuture 2026-04-28 11:43:29 +08:00
• c1d1412333 [Flashinfer] Integrate flashinfer router gemm for sm103 (#23285) Baizhou Zhang 2026-04-27 20:37:56 -07:00
• 3066ba8167 fix(hicache): add retry logic for MooncakeStore warmup (#17195) 3 people 2026-04-28 09:50:43 +08:00
• 3177fa7951 Enable DeepGemm warmup in DeepSeek-V4 cookbook (#23883) Baizhou Zhang 2026-04-27 18:41:12 -07:00
• b1e1fe8eee 【NPU】【bugfix】accuracy fix when enable both nsa cp and prefixcache (#23268) cen121212 2026-04-28 09:08:28 +08:00
• 9ffc0cc67e [NPU] Support GLM-4.5V (#22961) 看海的人 2026-04-28 09:08:19 +08:00
• 4a04a9818e 【hicache】Optimize HiCache prefetch logic: adapt to remaining available memory when memory is insufficient (#16370) PiteXChenandZhiqiang Xie 2026-04-28 08:18:35 +08:00
• 27659ea8c8 [PD+Pause] Remove redundant post processing (#23886) Byron Hsuandroot 2026-04-27 17:01:26 -07:00
• cb0429f253 [Disagg] Finalize routed_experts_output in process_batch_result_disagg_prefill (#23885) 3 people 2026-04-27 16:40:05 -07:00
• 7b9ff79f93 docs: update Python prerequisite to 3.10 (#23801) Pai Liu 2026-04-28 08:36:38 +10:00
• b73c44b545 test: relax TestMLADeepseekV3.test_gsm8k threshold 0.62 -> 0.60 (#23879) Alison Shao 2026-04-27 15:27:15 -07:00
• 41181b6238 [sgl] copy mm_input in piecewise cuda graph when eagle3 is on (#23613) Bi Xue 2026-04-27 13:35:19 -07:00
• f34c20af86 [VLM] Fix Kimi-K2.5 CPU path: rename grid_thws -> image_grid_thw (#23501) Kurt ShusterandEthan Su 2026-04-27 16:34:28 -04:00
• 28ee08c172 [HiCache] Add synchronization for context parallelism (#20460) Vladislav Nosivskoy 2026-04-27 21:13:07 +03:00
• f34222da1b [Docs] add cookbook for MiMo-V2.5 family (#23851) Xinyuan Tong 2026-04-27 18:38:41 +01:00
• 96b0c64c88 Add docs_new code owner (#23855) Xinyuan Tong 2026-04-27 18:31:35 +01:00
• 4fc5ebf0b7 [Chore] Remove deadcode in prefill delayer (#23389) Jonah BernardandJonah Bernard 2026-04-27 12:59:36 -04:00
• 47b8eadbc4 [Docs] Update Ascend NPU GGUF quantization documentation (#23845) 1874. 2026-04-27 22:30:24 +08:00
• f2b84b90ac [npu]fix: qwen3-next w8a8 precision bugs (#21698) ranjiewen 2026-04-27 18:14:33 +08:00
• 8536d4b402 Clean up noisy startup warnings from third-party deps (#23669) Lianmin Zheng 2026-04-27 03:10:46 -07:00
• 06725ecf0d [NPU] [DOC] Add support new models doc for NPU (#23824) amote-i 2026-04-27 17:13:23 +08:00
• a3fc982ba7 [Whisper] Automatic language detection via structured generation (#22997) Shenxiu LiuandXinyuan Tong 2026-04-27 00:54:41 -07:00
• c2ec64f243 docs: verify GB300 Pro DeepSeek V4 recipes (#23817) zijiexia 2026-04-27 00:21:26 -07:00
• 5f47cae1a0 add H100 configs for GLM-4.7-Flash (#23719) Xiaoyu Zhang 2026-04-27 15:07:39 +08:00
• 9ec4aa1b5c [Doc]Add msprobe doc in docs_new path (#23712) Yujing 2026-04-27 15:06:50 +08:00
• d49561b8ae [AMD] Fix Kimi-K2.6 Quark MXFP4 loading prefix and packed module mapping (#23408) Colin Z 2026-04-26 23:56:15 -07:00
• b7113cadb1 [Bug Fix] Reject pp_max_micro_batch_size=0 to prevent silent deadlock on generate() (#23799) Praneth Paruchuri 2026-04-27 11:06:04 +05:30
• e5198386bd Upgrade transformers from 5.5.4 to 5.6.0 (#23525) Xinyuan Tong 2026-04-27 06:33:54 +01:00
• 91825b8808 [FEAT][EPD] support encoder real health (#23343) Zheng Wengang 2026-04-27 13:21:28 +08:00
• 5141d8ae21 [AMD]fix: use CUDA event for targeted draft-to-verify sync in EAGLE overlap (#21940) AMD-yanfeiwang 2026-04-27 12:58:34 +08:00
• d84470079d [AMD] Fix Grok-2 nightly: avoid multimodal misdetection from auto-populated vision_config (#23383) Bingxu Chen 2026-04-27 12:54:36 +08:00
• c74b345e3a [docs] Update FA4 support SWA (#23793) zijiexia 2026-04-26 20:40:36 -07:00
• bead2e3470 perf: optimize PCG inductor path for FP8 models (redo of #21734) (#23227) Jia GuoandClaude Opus 4.7 2026-04-26 20:34:27 -07:00
• 85376a6119 refactor(moe): centralize post-experts all-reduce skip predicate (#23748) 3 people 2026-04-26 20:29:59 -07:00
• da175b964d chore: update CI test est_time values (#23785) sglang-botandsglang-bot 2026-04-26 20:17:50 -07:00
• 32c3513816 [NPU] Support MTP for Qwen3.5 (#20918) iridiumine 2026-04-27 10:44:17 +08:00