Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• ef361de916 [Model Runer V2][DFlash] Fix lm head sharing for dflash (#46435) Giancarlo Delfin 2026-06-23 14:09:06 -05:00
• acce57d8dd Deprecate old FP8 online MoE quantization class (#44514) Yan Ma 2026-06-24 02:53:38 +08:00
• 68afd78897 [Bugfix][ROCm] Fix cumem sleep and teardown (#46203) 4 people 2026-06-23 11:45:31 -07:00
• 37a682d392 [Kernel] Extend Marlin thread-tile padding to MoE (WNA16 + FP8/MXFP8) (#45703) Michael GoinandClaude 2026-06-23 12:45:10 -06:00
• d8e422ccda [Bugfix] Parse MiniMax M3 streaming reasoning by text markers (#45718) Rui Yin 2026-06-24 02:43:58 +08:00
• e368415daa [AMD][OCP MX][CI] Fix tests to not dispatch on UNFUSED_TRITON backend on MI300, improve w_mxfp4_a_fp8 emulation support (#46142) fxmarty-amd 2026-06-23 20:25:27 +02:00
• ceae5bcbda [ROCm][CI] Fix nixl tests (#45219) Andreas Karatzas 2026-06-23 13:11:40 -05:00
• 6691f087a6 [Minimax-M3] BF16/FP8 Indexer using MSA (#45892) 3 people 2026-06-23 10:28:49 -07:00
• f4d5f73ffa [Bugfix]: Fix unquantized gpt-oss weight loading broken by FusedMoE r… (#45818) Priyansh JainandLi, Jiang <jiang1.li@intel.com> 2026-06-23 22:26:17 +05:30
• fd50a66015 [CI][ROCm] Skip unsupported test cases on ROCm (#46160) fxmarty-amdandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-23 18:35:49 +02:00
• 84586c9acc [ROCm][CI] fix fp8 range in vit_fp8_quant (#46410) 3 people 2026-06-23 12:34:21 -04:00
• 40e5522121 [Docs] Add Qwen3 forced alignment online example (#46197) Taneem Ibrahim 2026-06-23 11:59:45 -04:00
• f3410b3bb1 fix(moe_wna16): access tp_size via moe_config for RoutedExperts compatibility (#45404) Willow Lopez 2026-06-23 23:46:23 +08:00
• 568874fec2 [ROCm][CI] pass merge-base to container for python-only wheel metadata (#45869) Divakar VermaandAndreas Karatzas 2026-06-23 11:44:43 -04:00
• 275b43183c [MyPy] Fix mypy for vllm/benchmarks (#39896) Martin HickeyandHarry Mellor 2026-06-23 16:22:29 +01:00
• 547d2c40d7 Add weights padding for fp8 per-block online quantization (#44763) Yan Ma 2026-06-23 23:08:17 +08:00
• 2aaaf3febd [ROCm][Test] Fix stale test_gfx950_moe MXFP4 oracle tests (#46260) 3 people 2026-06-23 08:07:46 -07:00
• 156b12667c [ROCm][CI] Skip Quark mxfp4 tests unless Quark version is compatible with Torch version (#46431) Micah Williamson 2026-06-23 09:26:39 -05:00
• 9f6f296428 [CI/Build] Remove BaiChuanForCausalLM from the LoRA test (#46494) Jee Jee Li 2026-06-23 22:09:48 +08:00
• e51e700470 [LoRA] Gate all_gather on fully_sharded_loras inside _mcp_apply; rewrite regression test (#45715) 3 people 2026-06-23 22:08:33 +08:00
• f59db63732 [Bugfix] GPT-OSS Autodrop reasoning in Response API and cleanup (#45048) yzong-rhandBen Browning 2026-06-23 09:36:33 -04:00
• 9f5117820f [HARDWARE][POWER] Enable fp16 support for PowerPC (#46135) Rukhaiya2004 2026-06-23 18:54:49 +05:30
• 1bf149f334 Filter Pydantic-internal markers from validation error param (#46457) Muhammad FawazandMahad Rehmann 2026-06-23 18:20:50 +05:00
• 2a675a7b9f [Bugfix] Responses API assistant EasyInputMessageParam input (#44361) yzong-rhandBen Browning 2026-06-23 08:54:46 -04:00
• 7d47cff933 [Bugfix][KV Offload] Fix swap_blocks_batch on the default stream (#46379) 3 people 2026-06-23 15:45:27 +03:00
• 091bc1026e [KV Offloading] Add tiering metric plumbing (#45959) Srinivas Krovvidiandsrinivas_oo7 2026-06-23 07:10:36 -05:00
• 3554ada5d8 [CPU][Bugfix][Speculative Decoding] Accept USE_FP64_GUMBEL in CPU recovered-tokens sampler (#46069) hillel.darshanandCursor 2026-06-23 14:54:07 +03:00
• 31ca9504b1 [Frontend] Split ServingRender into renderer and entrypoint. (#44285) wang.yuqi 2026-06-23 19:19:09 +08:00
• d32575a2d2 [ROCm][P/D] Support MoRIIO heterogeneous TP fan-in (#46332) 8 people 2026-06-23 18:33:23 +08:00
• 83fa302ca4 fix(security): prevent infinite loop in split_audio with NaN audio sa… (#46463) Juan Pérez de Algaba 2026-06-23 12:24:51 +02:00
• 20b5af55c1 [ROCm][Perf] DSv3.2: fuse MLA Q concat+fp8-quant in forward_mqa (#43673) frida-andersson 2026-06-23 12:12:04 +02:00
• 901a3b091c fix gpt_oss pp>1 with ep (#46441) Qiming Zhang 2026-06-23 01:59:11 -07:00
• 2d721ab5d8 [Rust Frontend] Align Rust allowed_token_ids validation with Python (#46348) ReidandBugen Zhao 2026-06-23 16:32:33 +08:00
• accaa434f3 [Rust Frontend] Support echo for token-ID completion prompts (#46219) Reidandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-23 16:04:41 +08:00
• a04654da23 Doc: fix missing GLM-5.x in supported models (#46452) Sunny Yuan 2026-06-23 00:42:27 -07:00
• 25bc3be49c [Rust Frontend] Correct --reasoning-parser semantics (#46359) Bugen Zhao 2026-06-23 15:38:39 +08:00
• a46f3eb232 [Bugfix][Model Runner V2] Preserve all allowed_token_ids in the logit bias kernel (#46245) Ting SUN 2026-06-23 14:01:13 +07:00
• 6c427dd401 [BugFix] Omit empty tool_calls from OpenAI chat responses (#44105) 3 people 2026-06-23 13:43:53 +08:00
• 3ce5823762 [Refactor] Responses API parser state into conversation context (#46030) Chaunceyandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-23 13:42:58 +08:00
• 04c2a8deac [DeepEP V2] Fill invalid recv_topk_idx with -1 (#46432) Woosuk Kwon 2026-06-22 21:45:49 -07:00
• 7e47fb72b5 [ROCm][P/D] Fix MoRIIO WRITE mode for mixed KV layouts (#46290) 7 people 2026-06-23 12:12:51 +08:00
• a8481be7a9 [Rust Frontend][Perf] Use dedicated runtime for HTTP/request-processing/ZMQ (#46051) Bugen Zhaoandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-23 12:03:20 +08:00
• 9d3317172c [XPU][CI]fix xpu kv cache layout test (#46429) Kunshang Ji 2026-06-23 11:43:29 +08:00
• 430a95ae3a [v1][kvcache] Honor prefix-cache retention interval for Mamba/linear attention (#45845) Dao007foreverandClaude Opus 4.8 2026-06-22 19:51:11 -07:00
• 56e5797511 [Quant] Enable modelopt_mixed on Turing (SM75) (#45375) Mike G 2026-06-22 19:30:49 -07:00
• 8db12169a4 fix: stream Qwen3 tool call string arguments (#46351) Rui Yinandabinggo 2026-06-23 10:26:37 +08:00
• 33f50773cb [Doc] Fix typos, grammar, and broken commands across docs (#46398) MichaelCaooandClaude 2026-06-23 10:01:22 +08:00
• fa36f86d77 [CI] Torch 2.11 flaky test_spec_decode_logprobs and gritlm tests (#45772) Micah Williamson 2026-06-22 20:26:54 -05:00
• 8207ce0850 [Bugfix] Fix humming lm_head crash and FusedMoE weight_shape coercion (#46420) Michael GoinandClaude Opus 4.8 2026-06-22 19:19:29 -06:00
• e48592066e [DeepEP V2] Bound num_max_tokens_per_rank in do_expand=False (#46404) 5 people 2026-06-22 18:14:53 -07:00
• 91ba720b75 [ROCm][CI] Only require q_scale==1.0 for fp8 query in RocmAttention (#46148) stefankoncarevicandAndreas Karatzas 2026-06-23 01:25:43 +02:00
• 6ead164e52 [CI] Add TP=4 requirement to test_mixed_precision_model_accuracies (#46161) fxmarty-amd 2026-06-23 01:19:43 +02:00
• c97e8f99d6 [ROCm][Quantization][4/N] refactor quark_moe fp8 w/ oracle (#43721) Bowen BaoandAndreas Karatzas 2026-06-22 15:58:03 -07:00
• 183b5f27ea [Bugfix][V1][TurboQuant] Reserve workspace before CUDA graph capture (#44053) 3 people 2026-06-23 06:47:48 +08:00
• ca5b24695b Fix static actorder handling for compressed-tensors WNA16 MoE (#41161) ZewenShen-Cohereandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-22 18:46:46 -04:00
• 6f6bd3b8fe [ROCm][CI] Increase the max wait time for server startup (#46417) Charlie Fu 2026-06-22 17:46:31 -05:00
• 70ef4d3009 [ROCm][CI] Purging away redundant test group definitions (#46418) Andreas Karatzas 2026-06-22 17:42:47 -05:00
• e2fe837572 [CI] Fix CPU-Multi-Modal Model Tests timeout by adding a 4th shard (#46388) Tyler Michael SmithandClaude 2026-06-22 18:08:00 -04:00
• fbf9ff7cf4 [CI][ROCm] Restrict MLA cross-layer KV cache test to supported backends on ROCm (#46401) Aarushi Jain 2026-06-22 17:05:26 -05:00
• 6cc2c9ba3a [CI] Add DGX Spark GPQA smoke test (#39541) Michael GoinandClaude Opus 4.8 2026-06-22 22:52:38 +01:00
• c0b2d8f471 [Bugfix] FusedMoE: coerce shape-(1,) per-tensor scales to 0-D scalar … (#43362) 3 people 2026-06-22 15:26:53 -05:00
• d1a38c2762 [Kernel][Performance] Add FlashInfer cutedsl NVFP4 GEMM backend (#42235) Mohammad Miadh Angkad 2026-06-23 04:17:18 +08:00
• 2b4a7491ec [ROCm][CI] Query total device memory via amdsmi to avoid HIP init (#46141) stefankoncarevicandAndreas Karatzas 2026-06-22 22:12:24 +02:00
• 82ede09a5a [Bugfix][KVConnector] Fix SimpleCPUOffloadConnector GPU->CPU store race (#46278) Saddss 2026-06-23 04:08:47 +08:00
• fbf520cf3a [MRV2] Generalize use of WhisperModelState (#46096) Nick Hill 2026-06-22 12:40:02 -07:00
• 44d95069e9 Enable DeepSeek V4 and GLM-5.1 on SM120 (#43477) 3 people 2026-06-23 02:54:14 +08:00
• 3ce15fd574 [v1][kvconnector] DecodeBenchConnector: fill list/tuple (Mamba/KDA) KV caches (#45080) Dao007foreverandClaude Opus 4.8 2026-06-22 11:54:00 -07:00
• e4b3da3feb [Quantization][CI] add humming lm-eval test (#43752) Jinzhen Linandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-23 02:23:55 +08:00
• 3e6529cc0e [Bugfix][Spec Decode] Fix EAGLE drafter multimodal encoder cache misses (#46315) 3 people 2026-06-22 11:14:02 -07:00
• ac614587f5 [EPLB] Enable nixl eplb communicator for elastic ep (#45013) Ilya MarkovandMarkov Ilya 2026-06-22 19:54:08 +02:00
• f2069b005b [Pooling] Validate non-negative rerank top_n (#46119) Taneem Ibrahimandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-22 11:40:47 -04:00
• ccd49f6821 [MyPy] Fix mypy for vllm/lora (#41722) Martin Hickey 2026-06-22 15:57:09 +01:00
• 1c7bc18318 [Bugfix][CPU] Fix CPU model runner v2 (#46365) Li, Jiang 2026-06-22 22:52:05 +08:00
• 9a938df64e [Test][KV Offloading] Add unit tests for OffloadingSpecFactory and SecondaryTierFactory (#46355) AlexHuang 2026-06-22 22:45:04 +08:00
• 3da4a1b124 [XPU] add awq format for INCXPULinear (#43404) Liangliang Ma 2026-06-22 22:29:13 +08:00
• 6871738777 [Doc] Document pull request limit (#46376) Simon MoandOpenAI Codex 2026-06-22 07:04:56 -07:00
• aa4990a9a2 [Attention] Re-enable cross-layer KV cache layout for MLA via stride-aware kernels (#45111) Yifan Qiao 2026-06-22 06:57:02 -07:00
• a4610da0c6 [docs] link security docs from AGENTS (#46373) Simon MoandOpenAI Codex 2026-06-22 06:28:25 -07:00
• 09cdcf34aa [XPU] update nixl to v1.2.0 (#46327) liuzhenwei 2026-06-22 20:55:06 +08:00
• d2c671c29b [CPU][RISC-V] Add RVV micro GEMM for WNA16 (#44324) wcyandLi, Jiang <jiang1.li@intel.com> 2026-06-22 20:53:54 +08:00
• b5a2adec4b [XPU][CI]Skip v1/spec_decode/test_speculators_correctness.py in intel GPU nightly (#46356) xiangdong 2026-06-22 19:30:41 +08:00
• 78739e3bda [Bugfix] Reject matryoshka embedding dimensions above hidden size (#46313) Maxwill LinandEazyReal 2026-06-22 03:16:35 -07:00
• 89accad2cc [ROCm][DSV4] Disable TileLang MHC dispatch on gfx942 (#45931) Tuukka Sarvi 2026-06-22 12:26:54 +03:00
• 3c8e49596c [Model] ColQwen3.5: fix retrieval correctness (bias + bidirectional) (#46108) Athrael SojuandClaude Opus 4.8 2026-06-22 10:25:54 +01:00
• cec2ec1176 [Bugfix] Avoid racy accepted counts in async spec decode (#45100) 3 people 2026-06-22 01:53:16 -07:00
• 435f82d61a [Bugfix] Fix Llama4ForCausalLM initialization test failure (#46341) liuzhenwei 2026-06-22 16:40:43 +08:00
• 1c4b51b990 Temporarily skip M3 on CI (#46352) Roger Wang 2026-06-22 01:35:31 -07:00
• 2e2c47928b [Doc] Update MiniMax-M3 (#45940) 3 people 2026-06-22 16:23:27 +08:00
• 80abe0de7d [Rust Frontend] Support thinking_token_budget for chat and completions (#46137) 3 people 2026-06-22 16:00:02 +08:00
• a9f7b2d41c [feature][kv_offload] Self-describing KV events for OffloadingConnector (#43468) Change72andClaude 2026-06-22 00:27:46 -07:00
• d14e551a53 [Model] Remove MiniMaxText01, MiniMaxVL01, MiniMaxForCausalLM (#45993) 3 people 2026-06-22 15:20:46 +08:00
• 68567ef2df [CPUOffloadingManager] Maintain evictable list in LRUCachePolicy (#46216) Varun Sundar RabindranathandVarun Sundar Rabindranath 2026-06-22 02:54:44 -04:00
• 6bc6f2d86d [1/N][Core] add partial prefix cache primitives (#45939) Jiangyun ZhuandYifan Qiao 2026-06-22 14:43:10 +08:00
• 1eb2cc961e [Frontend] Refactor ServingTokenization entrypoint. (#46022) wang.yuqi 2026-06-22 14:27:58 +08:00
• 31124749d1 [Bugfix] [Rust Frontend] Fix stop string truncation with repeated matches (#46113) Reidandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-22 14:11:29 +08:00
• 9037498c22 [DSV4][XPU] Pass gemm1_clamp_limit to XpuFusedMoe (#44517) Ma Jian 2026-06-22 12:57:10 +08:00
• db32b53e30 [SpecDecode] Support DFlash with FlashInfer (#43081) gq112andgss 2026-06-22 12:55:30 +08:00
• b529bfd6c5 [XPU][CI] Add agent_tags for Intel GPU CI (#45768) xiangdong 2026-06-22 10:33:17 +08:00
• f3df7a7231 [ROCm][CI] Enable kv_connector unit tests on ROCm (#45955) Micah Williamson 2026-06-21 21:08:44 -05:00
• 485bbe1c6f [CI] Fix missing tp_size attribute on RoutedExperts (#46163) fxmarty-amdandAndreas Karatzas 2026-06-22 02:46:49 +02:00