Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• 5b3762a7f0 [Bugfix][CPU] Fix Clang OpenMP build on macOS (#49021) SyaandLi, Jiang <jiang1.li@intel.com> 2026-07-21 17:58:52 +08:00
• 4d30c510ce [bugfix] Fix Cosmos3 Edge checkpoint weights filtering, video loading, prompt expansion (#49190) bastefaniak 2026-07-21 11:18:36 +02:00
• 6700813f86 [3/N][KV-Cache Layout Refactor] Standardize Mamba cache; drop get_transfer_cache_regions (#44456) 3 people 2026-07-21 11:16:15 +02:00
• eb44b3aaa4 [Rust][Benchmark] Use async HTTP clients (#49295) Bugen Zhao 2026-07-21 16:53:57 +08:00
• 7a98c7a392 [Misc] Remove old now unsupported max_num_partial_prefills and max_long_partial_prefills (#49244) Nicolò Lucchesi 2026-07-21 10:52:52 +02:00
• 0d9e60619b [Misc][Docs] Fix XPU compute-runtime driver link version mismatch (#49299) Lena Onyshchenko 2026-07-21 10:45:41 +02:00
• 1134545b6f Revert "[Sampler] Stop upcasting logits to fp32 in apply_sampling_params" (#48641) (#49033) vllm-agentandvllm-agent 2026-07-21 01:36:45 -07:00
• 3e0c887511 [Bugfix] Fix Ovis2_5 special tokens for transformers v5 (#47298) 3 people 2026-07-21 10:09:20 +02:00
• adfbbc1005 Propagate Flash Attention cache configuration to Ray workers (#49177) Stefan Kaestle 2026-07-21 09:47:53 +02:00
• adc98f04d0 [Misc] Add @esmeetu to codeowners for rust/src/bench (#49298) Roy Wang 2026-07-21 15:44:24 +08:00
• 8def3cdde2 [Bugfix] Propagate quant_config to LFM2 ShortConv projections (#48917) Alex YuanandClaude Opus 4.8 2026-07-20 23:51:55 -07:00
• 616c9bd0f4 [Frontend] Support additional sampling parameters for translation API (#45839) Guan-Ming ChiuandCyrus Leung 2026-07-21 13:53:56 +08:00
• 8688a06d67 [Rust][Benchmark] Use tracing for logs (#48937) Bugen Zhao 2026-07-21 13:18:13 +08:00
• f25953cc59 [Bugfix][Rust Frontend] Handle zero-column logprobs payloads without panicking (#49113) Salt Satoandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-21 05:30:29 +01:00
• d9aa35161d Update BGE-M3 token expectations for leading spaces (#49269) 3 people 2026-07-21 11:49:17 +08:00
• 6bcda970fd [CI][NIXL] Isolate concurrent engine internal ports (#49129) Andreas KaratzasandOpenAI Codex 2026-07-20 22:28:11 -05:00
• ea0e9c8f2e [MRV2] Add encoder cache profiling implementation (#47985) Isotr0py 2026-07-21 11:18:14 +08:00
• 94ed0bf4e0 [Bugfix][KV Offloading] Handle queued request aborts without allocated KV blocks (#49146) Chauncey 2026-07-21 11:16:26 +08:00
• 1940c8441e [Rust Frontend][gRPC] Add engine-aware health reporting (#48992) Connor Carpenterandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-20 19:54:39 -07:00
• 72d16aee15 [CI] Exercise FA3 FP8 attention on SM90 (#49231) Simon Mo 2026-07-20 19:26:55 -07:00
• e78a0c8e59 [XPU][Doc] Update XPU docker image documents (#49148) Kunshang Ji 2026-07-21 10:11:27 +08:00
• 97a98006b0 Update qutlass cmake for stable abi (#47879) Chris Leonard 2026-07-20 21:31:10 -04:00
• 0a684ab0c0 [Bugfix] Fix WSL circular import from pin_memory warning_once (#48444) Alejandro Paredes La TorreandShengqi Chen 2026-07-20 18:30:55 -07:00
• 0d9210a502 Fixes non-coalesced HBM access in marlin_int4_fp8_preprocess_kernel_awq (#47268) 3 people 2026-07-21 09:30:38 +08:00
• 1d874867ea [Misc][Docs] Fix broken protocol link in speech_to_text doc (#47212) Lena OnyshchenkoandCursor 2026-07-21 03:05:05 +02:00
• 2e2e626b40 [Bugfix] Count per-group blocks in get_max_concurrency_for_kv_cache_config (#48317) 5 people 2026-07-20 19:29:10 -05:00
• af91f4b3e4 [Cleanup] Remove unused StructuredOutputRequest.status field (#49235) Nick Hill 2026-07-21 01:06:16 +01:00
• 2396a61108 [Attention][MLA][DCP] Query replication for MLA decode (DeepSeek-V2/R1 + Kimi-K2.5) (#45964) 4 people 2026-07-20 16:51:27 -07:00
• 97a668152b [RL Infra][FlashInfer] Enable router replay output from FlashInfer monolithic MoE kernel (#44214) 4 people 2026-07-21 01:45:10 +02:00
• 58b2012aa2 [copy of #45208] CuMem slept-L1 fragmentation accounting (#49208) 4 people 2026-07-20 19:08:11 -04:00
• b7c20d0cfa [chore] adjust logo be more friendly to white background terminal (#48938) Ning Xie 2026-07-21 06:15:28 +08:00
• a2b1f9fc3b [ROCm] [Release] [Bugfix] Fix the per commit wheel release pipeline. (#49245) TJian 2026-07-20 15:12:38 -07:00
• 642076d26c Support loading sample_from_anchor flag from speculators config (#48639) Fynn Schmitt-UlmsandMichael Goin 2026-07-20 17:52:43 -04:00
• 5feb3950e5 [ROCm][CI] fix test_rocm_quick_reduce.py (#49234) Charlie Fu 2026-07-20 16:39:33 -05:00
• 4ec199b66a [Bugfix][Spec-Decode] Populate draft seq_lens_cpu_upper_bound for spec-decode attention metadata (#44492) 4 people 2026-07-20 14:50:58 -06:00
• 7ca017778f [Feat][Perf] Add new warmup infrastructure for JITs (#47451) Roberto L. Castroandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-20 22:21:55 +02:00
• fbfe58133d [Bugfix][KV Offload] Preserve reachable tails for hybrid SWA groups (#48911) 3 people 2026-07-20 13:12:36 -06:00
• 9dd62d80ab Cosmos3 FP8 ModelOpt/Diffusers remapping (#48952) 3 people 2026-07-20 20:31:48 +02:00
• f878367898 [Revert][Bugfix] Restore MiniCPM-V 4.6 ViT QKV weight loader (#49193) tc-mbandwjinxu 2026-07-21 02:17:46 +08:00
• bd091079cb [Attention] FlashAttention 4 SM100 FP8 kv cache support (#42569) Matthew Bonanniandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-20 13:53:27 -04:00
• b23bd73f54 [XPU]add sycl path for Mhc (#47245) LiangqiusongandKunshang Ji 2026-07-20 23:32:54 +08:00
• e2d7adeb64 [Rust Frontend] Bump xgrammar-structural-tag and enable local extension (#49161) Bugen Zhao 2026-07-20 23:22:24 +08:00
• 15cb8e140d [Multimodal] Allow keeping original image mode for ImageIO (#49159) Isotr0py 2026-07-20 21:42:45 +08:00
• f007cceb42 [KV Offload] Support self-describing KV events with TieringOffloadingSpec (#48679) Chang GuoandOpenAI Codex 2026-07-20 06:41:58 -07:00
• 0a5069e4e3 [Bugfix][Gemma4] Fix ModelOpt mixed-precision MoE config mapping (#48563) wangqianandOpenAI Codex 2026-07-20 21:39:28 +08:00
• 8ce53a616e [Bugfix] Zero new KV blocks for quantized + sliding-window hybrid caches (#47574) 3 people 2026-07-20 09:18:17 -04:00
• ae10e855ab [Misc][Docs] Remove duplicate CodeGeex4 row in XPU model table (#47210) Lena Onyshchenko 2026-07-20 13:05:36 +03:00
• 530ee36a0d fix(openai): reject non-numeric logprobs with 400 instead of 500 (#49144) hcl 2026-07-20 18:04:50 +08:00
• d835ad572c [Bugfix][Rust Frontend] Map missing prompt logprobs for single-token prompts in chat and raw generate (#49111) Salt Sato 2026-07-20 11:00:06 +01:00
• 47d0597ca2 [Misc][Docs] Fix broken csrc kernel links in fusions doc (#47211) Lena OnyshchenkoandCursor 2026-07-20 12:44:25 +03:00
• 818cf61e91 [Rust Frontend] Fix macro-based content format detection (#49042) Reid 2026-07-20 17:39:13 +08:00
• c01618fdc8 [Rust][Benchmark] Integrate vllm-bench to vllm-rs & vllm CLI (#48930) Bugen Zhaoandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-20 17:31:25 +08:00
• 823eaf667d [XPU] FP8 o_proj with fp8_bmm and load-time scale transpose (#48334) 3 people 2026-07-20 16:32:03 +08:00
• f1f1259692 [Rust Frontend] Use zero-copy slicing for multimodal tensors (#48781) Sageandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-20 11:28:25 +03:00
• df13b5aef5 [XPU] [MoE] add quant input when prepare for fusedmoe (#47122) 4 people 2026-07-20 15:47:26 +08:00
• 4938d44a3b [CPU] fixes heterogeneous NIXL KV transfer into CPU_ATTN decode workers (#47871) Sihan ChenandLi, Jiang <jiang1.li@intel.com> 2026-07-20 15:33:13 +08:00
• 37bf988c2f [XPU][Bugfix] Fix GroupCoordinator device_index (#47295) Michał GanczarenkoandKunshang Ji 2026-07-20 09:25:56 +02:00
• 9459fc6471 [Bugfix][RL] Set vLLM config during weight reload (#45989) aoshen02 2026-07-20 15:02:56 +08:00
• 5245c80564 [Doc] Document blocks_per_chunk in the KV offloading guide (#49100) Itay EtelisandItay Etelis 2026-07-20 09:48:43 +03:00
• 9bc266d923 [Bugfix][KV Offload] Propagate EAGLE mode to SimpleCPU coordinator (#49071) Yifan QiaoandOpenAI Codex 2026-07-19 23:39:11 -07:00
• 5c9f6557d7 [Hardware][CPU] Enable granite-4 model on cpu (#47641) 6 people 2026-07-20 11:45:16 +05:30
• dcfebf93f4 [Bugfix] Fix logprobs token-string collision from SentencePiece space… (#48674) 4 people 2026-07-20 12:17:18 +08:00
• 752bd10647 [ROCm][CI] Fix sparse MLA metadata sync fixture (#49128) Andreas KaratzasandOpenAI Codex 2026-07-19 23:02:03 -05:00
• 2730b657c4 [Bugfix] Fix broken NVVM caused by CuteDSL 4.6.0 (#49108) Thien Tran 2026-07-20 10:45:56 +08:00
• 1dcbbd9cac [CI] Move compatible 1xL4 jobs to H200 35GB MIG (#43024) 4 people 2026-07-19 19:21:25 -07:00
• ace9fda495 [CI/Build][BugFix][The Rock][AMD] Add spawn method in vision examples to avoid reinitialization (#47932) rasmithandAndreas Karatzas 2026-07-19 13:41:52 -05:00
• ef0aa7ca2f [ROCm] [Release] [Per-commit] Reenable per commit rocm wheel (#49044) TJian 2026-07-19 11:38:04 -07:00
• e6d1310b2a [Bugfix] Reject removed pooling parameters (#48984) Taneem Ibrahim 2026-07-19 07:18:03 -05:00
• ac5f38a0f7 [Refactor] Extract StructuredOutputsParams creation logic from Request.to_sampling_params (#49003) yzong-rh 2026-07-19 08:18:00 -04:00
• b6ff8a2f50 [Core] Add MRV2 virtual-batch PCP for MLA (#46570) 4 people 2026-07-19 04:53:15 +02:00
• 9243e0124e [Multimodal] Automatically fallback to ViT DP when TP is unavailable (#49046) Isotr0pyandRoger Wang 2026-07-19 05:41:04 +08:00
• df362b2d6d [ROCm][CI] Ensure sliding window tests release GPU memory (#49055) Andreas Karatzas 2026-07-18 15:44:05 -05:00
• 7c2acd38b7 [Bugfix] Qwen3-VL/Qwen-Omni: honor max_pixels/min_pixels for video prompts (#49015) SYLAR 2026-07-19 01:29:11 +08:00
• a287eb163f [Front-end] [Messages] Populate num_cache_creation_tokens (#48535) yzong-rh 2026-07-18 13:04:35 -04:00
• e94243893d [ROCm][DSv3.2][Perf] Cap sparse MLA decode KV-splits with a work-per-split heuristic (#46832) frida-andersson 2026-07-18 18:39:37 +02:00
• 29c0ec4d63 [ci] Move 3 entrypoints tests to h200_35gb queue (#43164) 3 people 2026-07-18 08:43:49 -07:00
• c7ce03bcbd [Bugfix] Bump tml-fa4 for cutlass-dsl 4.6 API compatibility (#48988) Michael Goin 2026-07-18 08:59:33 -04:00
• c233d90aa8 Remove even more unnecessary load_weights methods (#48496) Harry Mellor 2026-07-18 09:40:27 +01:00
• d96aee0951 [Bugfix] Re-sync parameter tp_rank after process_weights_after_loading (fix replicated / disable_tp weight reload) (#48025) 4 people 2026-07-18 01:40:06 -07:00
• c71a583aa9 [Perf][Hybrid] Vectorize _copy_mamba_state_block to uint64 for temporal (#48110) Francesco Fusco 2026-07-18 06:43:09 +02:00
• f12b80c6ef [ROCm][Bugfix] Fix GPT-OSS Quark MXFP4 MoE loading - emulation buffer not block-aligned (#43979) xuebwang-amd 2026-07-18 11:49:41 +08:00
• da64db78b9 [LoRA] Optimize TrtLlmLoRAExperts (#48759) Jee Jee Li 2026-07-18 10:26:14 +08:00
• 425c4eafb0 [Sampler] Stop upcasting logits to fp32 in apply_sampling_params (#48641) 3 people 2026-07-17 21:15:20 -04:00
• 02c01f442b [Model] Use standard ModelOpt config for Inkling NVFP4 (#48990) Michael Goin 2026-07-17 21:13:14 -04:00
• fae543015c [Frontend]Flatten beam-search beams with itertools.chain instead of sum (#48829) 3 people 2026-07-18 06:09:14 +08:00
• c9be3a8aa1 [Kernel][Helion] Disable warp specialization in rms_norm_per_block_quant B200 configs (#48797) Shangdi YuandClaude Opus 4.8 2026-07-17 14:39:09 -07:00
• 41ea2dd44a [Bugfix][V1/V2] Fix prompt_logprobs to respect logprobs_mode (#47680) 3 people 2026-07-18 04:58:59 +08:00
• 088c0be268 [CI] Fix macOS wheel release annotation context (#48771) Kevin H. LuuandCodex 2026-07-17 13:44:48 -07:00
• fcd2255d16 [Hardware][GPU] Profiler config additional to increase it scope and annotation details (#37524) devalshahamdandDeval Shah 2026-07-17 13:38:59 -07:00
• b5433b6f50 [Perf] Optimize dsv4 routing using specialized kernel, 2.94% E2E TPOT improvement (#48660) Wentao Ye 2026-07-17 16:35:06 -04:00
• cc25f028b7 [Loader] Improve InstantTensor loading (#46868) Michael GoinandOpenAI Codex 2026-07-17 16:30:02 -04:00
• c4cd2bd544 [Bugfix] MoRIIO toy P/D proxy: fix DP-rank index aliasing + harden for high-concurrency bursts (#46115) 4 people 2026-07-17 12:35:04 -07:00
• 5784507da4 [Attention] Allow selecting a different attention backend per KV-cache group (#48012) Nicolò LucchesiandClaude Opus 4.8 2026-07-17 21:19:02 +02:00
• bf578e1abd [Bugfix][GLM4V] Fix video dummy profiling and memory usage (#48729) labAxiaoming 2026-07-18 01:44:45 +08:00
• efed8a1e83 [ROCm][Perf][DSV4] Improve sparse decode reduction occupancy on gfx950 (#48788) Fangzhou Ai 2026-07-17 13:24:59 -04:00
• 11d291511a [Bugfix][Tool Parser] Preserve whitespace in parameter values (MiniMax M2, Qwen3, MiniCPM5 XML) (#48846) 3 people 2026-07-17 19:45:41 +03:00
• 877dae9c68 [Refactor] Remove deepseek dead code (#48780) Wentao Yeandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-17 10:57:13 -04:00
• c4dd6d78fd Fix: Restore data_parallel_size > 1 for use_sequence_parallel_moe (#48849) passtoor-agi 2026-07-17 17:27:50 +03:00
• ce2aecc4dc [Performance] Use CuTe-DSL for FlashInfer MXFP4 quantization (#48417) JooHo Lee 2026-07-17 22:53:48 +09:00
• f38f3d11fb [Bugfix][KV Offloading] Offload last block at request finish and prevent reuse race (#48596) AlexHuangandOr Ozeri 2026-07-17 21:50:49 +08:00