Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• 2ac125123a [CI] Use explicit devices in IR tests (#49513) Andreas Karatzas 2026-07-23 22:08:14 -05:00
• 7bdf8cc37c [Bugfix] Fix humming kernel crash when layer.has_bias is None (#48769) Kyle SayersandClaude Sonnet 4.6 2026-07-23 23:07:45 -04:00
• bf27e34ebb [CompressedTensors] DeepSeek4 CT Quantization Support (#41276) Kyle Sayers 2026-07-23 23:07:31 -04:00
• 275556c35c [Bugfix] Detect mixed precision in packed KV cache specs (#49623) Michael GoinandOpenAI Codex 2026-07-23 23:07:06 -04:00
• d65acd83d8 [Model] Support llm-compressor Inkling NVFP4 weights (#49258) Michael Goin 2026-07-23 23:06:01 -04:00
• 80c9d5d5e0 [ROCm][Quantization] Add Quark W4A8 (INT4-FP8) MoE CI coverage (#48050) 4 people 2026-07-23 19:56:40 -07:00
• da54a5bf05 [Docs] Fix broken anchor links in serving/pooling/MoE docs (#49654) Euisuh Jeong 2026-07-24 11:42:56 +09:00
• 1479bd9e9d [ROCm][CI] Prepare AMD mirrors for regating (#49270) Andreas KaratzasandOpenAI Codex 2026-07-23 20:42:25 -05:00
• 0231dd5467 [BugFix][LoRA] Skip marlin-backend gpt-oss LoRA tests on XPU (#49385) Chaojun Zhangandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-24 09:16:00 +08:00
• 2659467497 [CI][PD] Add hybrid SSM P_TP>D_TP accuracy sweep entry (#49593) Nicolò Lucchesiandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-23 23:49:53 +02:00
• a49d37c6b9 [CI] Disable reasoning in Responses smoke test (#49511) Andreas Karatzas 2026-07-23 14:02:54 -05:00
• 4501a6d56b [ROCm][CI] Language Models tests tiny-mixtral with aiter fix (#49551) music-dino 2026-07-23 21:01:36 +02:00
• e18f0037a5 [Bugfix][KV cache] Support sparse-MLA targets with SWA drafts (#48776) Michael GoinandOpenAI Codex 2026-07-23 14:30:50 -04:00
• b354734d17 [Bug] Fix batch invariance rms norm comparison (#49603) Wentao Ye 2026-07-23 14:27:38 -04:00
• b91a40e729 [Bugfix] Restore structured output logger initialization (#49626) Chang GuoandClaude Opus 4.8 2026-07-23 11:26:03 -07:00
• 75ccdf3145 [Core] Update PyTorch to 2.13.0, torchvision to 0.28.0, triton to 3.7.1 (#48155) 3 people 2026-07-23 14:09:36 -04:00
• c6fe94b4d5 [CI] Bump PyTorch Compilation Unit Tests timeout to 150 min (#49606) Andrey Talman 2026-07-23 14:08:45 -04:00
• 46f01a50ac [CI][Bugfix] Fix test isolation in block_int8/ptpc_fp8 MoE kernel tests (#49609) Nick HillandClaude Opus 4.8 2026-07-23 18:26:02 +01:00
• f00efc5265 [CI] Isolate cudagraph tests in child processes (#49510) Andreas Karatzas 2026-07-23 12:16:19 -05:00
• b0cb1da1bd [DSv4 Perf] Skip topk and router when not needed, 3.4% E2E TTFT improvement for Decode case (#49486) Wentao Ye 2026-07-23 13:08:08 -04:00
• 0e36e3bbd1 [CI] Use explicit devices in quantization tests (#49512) Andreas Karatzas 2026-07-23 11:54:26 -05:00
• 494845e79f Revert "[MRV2] Always build attn metadata at capture time" (#49364) (#49451) vllm-agentandvllm-agent CI bot 2026-07-23 09:51:33 -07:00
• 0416dab275 [Bugfix][Structured Output][Spec Decode] Advance grammar across reasoning boundary (#44993) yue.yu 2026-07-24 00:14:15 +08:00
• c8db00b16c Fix GPTQ quantized Qwen3.5 MTP weight loading with spec decode (#48816) 3 people 2026-07-23 21:59:48 +08:00
• 80c7683923 [Perf] Defer MM embeds loading off the event loop (#49477) Guan-Ming Chiu 2026-07-23 21:58:50 +08:00
• 638d6e9757 [Bugfix][CI/Build] Fix Plamo2 HF runner crash on transformers v5 (_tied_weights_keys list→dict) (#44239) 3 people 2026-07-23 05:25:14 -07:00
• 1ad84fea86 [Bugfix][Spec Decode] Select earliest-completing stop string in check_stop_strings (#49391) Junpu Yu 2026-07-23 03:14:06 -07:00
• 12213c6795 [Bugfix] handle grammar compilation failures to avoid engine crash (#47312) zhrrrandNick Hill 2026-07-23 18:13:49 +08:00
• 10c75477b0 [Bugfix][Core] shm_broadcast: bound idle reader waits and release read slots (#45224) 4 people 2026-07-23 19:13:29 +09:00
• ac36a7a1e7 [MRV2][Spec Decode] Avoid rejection sampler OOM by chunking (#48630) Michael GoinandNick Hill 2026-07-23 06:13:13 -04:00
• 521aa80f71 [Core] Simplify KVBlockZeroer index tensor handling (#48399) Nick Hillandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-23 11:12:46 +01:00
• a76df87db8 [MooncakeStore] Re-derive full external hits on stored boundaries (#49481) 3 people 2026-07-23 03:12:28 -07:00
• a4904ba903 [Perf][KVConnector][Mooncake] Vectorize prepare_value on the KV load path (#48531) 4 people 2026-07-23 03:12:00 -07:00
• f83de6d44c [CPU][Docs] Update docs and dockerfile for s390x (#49523) Rehan Khan 2026-07-23 12:47:24 +05:30
• 239fc73553 [Misc] Use VLLMValidationError in chat_utils content-part validation (#49217) Umut Polat 2026-07-23 08:48:37 +03:00
• 76bf55240c [Bugfix] Fix DeepSeek-V4 DSpark draft shared-expert padding for TP > 8 (#49415) Mike G 2026-07-22 22:06:21 -07:00
• 9a698f3255 [Performance][Model] Avoid transient Inkling result allocations (performance, and OOM prevention on smaller memory configurations) (#49487) 3 people 2026-07-22 22:03:34 -07:00
• 4080263bb2 [Bugfix][Model] Remove SciPy dependency from Inkling scale planning (#49485) 3 people 2026-07-22 21:34:57 -07:00
• fc5fda105f [Docs] Re-add Reo.dev analytics beacon (#49474) jcotant-inferact 2026-07-22 20:03:32 -07:00
• b07ec92faa [Bugfix] Make shared NVFP4 MoE scales writable (#49489) Matej Sirovatka 2026-07-22 19:17:53 -07:00
• 27ffbfde8d Fused Shared Expert Support for AMD Quark DeepSeek-V4 Model Checkpoints (#48044) Colin Zandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-22 17:34:17 -07:00
• 229e01e9e1 [BugFix] Handle per-group prefix-hit divergence for hybrid models with KV connector (#48425) Nick Hill 2026-07-23 01:19:11 +01:00
• 191146dba5 Add quantization label automation (#49492) Michael GoinandOpenAI Codex 2026-07-22 19:59:15 -04:00
• f3a920a076 [Core][DSV4] Compact MXFP4 indexer KV cache and packed group overlays (#48993) Summer Yang 2026-07-22 16:58:44 -07:00
• 149daf0d72 [Bugfix] Exclude location-derived path vars from torch.compile cache factors (#47573) Nils Mattesonandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-22 17:56:09 -06:00
• 917fdb5bf7 [Bugfix] Fix DeepGEMM warmup when using FlashInferFp8DeepGEMMDynamicBlockScaledKernel (#49467) Michael Goin 2026-07-22 19:28:49 -04:00
• 4b594b4aa1 [Bugfix][CI] Fix topk_softplus_sqrt no-op on non-XPU platforms (#49452) stefankoncarevic 2026-07-23 00:36:17 +02:00
• 7d10a4cfce [Bugfix] Retry config read to survive concurrent HF cache refresh (#49001) peizhang56andClaude 2026-07-22 15:35:38 -07:00
• 910cc8543a [Bugfix] Restore gather_and_maybe_dequant_cache OOB guard (#49427) Nick Hill 2026-07-22 21:47:37 +01:00
• 431934522b [CI] Fix stale/fragile untethered kernels-root tests (#49423) Nick Hill 2026-07-22 21:43:07 +01:00
• 61a09532f2 Bump Flashinfer version to 0.6.15 (#48914) 3 people 2026-07-22 16:32:05 -04:00
• 3de4b2bf3c [Bugfix][Parser] Fix special tokens (EOS/BOS) leaking into reasoning content (#48748) Ben Browningandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-22 16:10:55 -04:00
• b44311b6ef [CI] stabilize GDN prefill CuTeDSL test (#49388) Thien TranandCodex 2026-07-23 00:03:46 +08:00
• b0d7875180 [CI] Increase timeout of pytorch-compilation-unit-tests (#49450) Nick Hill 2026-07-22 16:39:09 +01:00
• 53c2f20dd9 [ROCm][CI] skip moe weight padding for eplb (#49350) Divakar Verma 2026-07-22 11:18:01 -04:00
• 37e370fe93 [DSv4 Perf] Skip empty c128 kernel launch, around 2x kernel performance improvement. (#48957) Wentao Ye 2026-07-22 10:55:20 -04:00
• 2dc5a72e7e [Bugfix][Renderer] Rebuild vision chunk UUIDs in async render path (#49400) Guan-Ming Chiu 2026-07-22 22:11:24 +08:00
• c79ff5f918 [Build] Bump vllm-flash-attn to C++20-compatible commit for torch-nightly (#49326) Andrey Talman 2026-07-22 09:51:59 -04:00
• 1a659a0c37 Upgrade tpu-inference to v0.25.0 (#49431) Teresa Chen 2026-07-22 19:52:56 +08:00
• 0f6cf7f628 [Rust Frontend] Extract request preparation from the inference path (#49045) Sage 2026-07-22 14:31:36 +03:00
• c79ad3ae21 [Rust Frontend][gRPC] Add abort control RPC (#49255) Connor CarpenterandOpenAI Codex 2026-07-22 04:31:01 -07:00
• 61c9ef986a [Frontend] Parallelize preprocessing within the same request for pooling models online serving. (#49153) wang.yuqi 2026-07-22 18:56:23 +08:00
• d6dbdb9b0d [XPU] WA of topk_softplus_sqrt arg mismatch on XPU (#49408) Liangqiusong 2026-07-22 16:16:13 +08:00
• 06da482fb4 [XPU] WA of topk_softmax arg mismatch on XPU (#49395) liuzhenwei 2026-07-22 16:05:01 +08:00
• 2f75e7f712 [CI] Increase timeouts for jobs exceeding current limits (#49374) Kevin H. LuuandOpenAI Codex 2026-07-22 00:16:16 -07:00
• 7c21548ce3 [PD][Bugfix] Fix NIXL hybrid MLA+mamba heterogeneous TP (#49297) Ziming Huang 2026-07-22 15:11:39 +08:00
• 9df2f91232 [Renderer] Offload derender CPU work to renderer thread pool (#49396) Guan-Ming Chiu 2026-07-22 14:55:03 +08:00
• 387189c429 [ROCm] Remove redundant AITER fused_qk_rmsnorm probe (avoids config-time HIP init) (#47992) stefankoncarevicandAndreas Karatzas 2026-07-22 05:58:14 +02:00
• 75576c63be Add auto label for xpu relate issue (#49398) Kunshang Ji 2026-07-22 11:33:20 +08:00
• 16aca639b7 [ROCm] Upgrade NIXL and UCX (#49251) Andreas Karatzas 2026-07-21 22:30:02 -05:00
• 6049424b7e [MRV2] Always build attn metadata at capture time (#49364) Woosuk Kwon 2026-07-21 19:27:16 -07:00
• 060b5f61dc [Bugfix][Attention] Ignore empty MLA context chunks during merge (#49294) Lucas WilkinsonandOpenAI Codex 2026-07-22 04:23:48 +02:00
• ec59c1579f [MoE Refactor] Migrate MoeWNA16Method quantization method over to using the new MK oracle scheme. (#44120) 3 people 2026-07-21 22:20:30 -04:00
• 1750e443f2 [Misc] Move PyNvVideoCodec stuff out of gpu worker (#49322) Isotr0py 2026-07-22 10:11:56 +08:00
• ba18929079 [Bugfix][SpecDecode] Scope MTP completeness checks outside bucketed updates (#49178) aoshen02andaoshen02 2026-07-22 09:56:39 +08:00
• 0500ca6a58 [CI][Bugfix] Fix ROCm FP8 KV cache dtype in attention backend test (#49380) peizhang56andClaude 2026-07-21 18:34:49 -07:00
• a1c15bcb0f [CI][Bugfix] Fix and wire streaming-input tests (#49356) Nick HillandClaude Opus 4.8 2026-07-22 01:35:23 +01:00
• 4809de7317 [Misc] Fix terminal output logo coloring (#49344) Nick Hill 2026-07-22 01:21:29 +01:00
• 05781e21dd [ROCm][CI] Fix order-dependent failure in test_flash_attn_accepts_handled_fp8_variants (MI355) (#49329) stefankoncarevic 2026-07-22 01:54:33 +02:00
• 85f638a2b8 skip cudagraph/DP padding in topk (#48979) gnovack 2026-07-21 15:20:30 -07:00
• 08e5067561 [CI] Bump timeout of entrypoints-integration-api-server-openai-part-2 (#49359) Nick Hill 2026-07-21 22:54:03 +01:00
• a7d00ec051 [Bugfix] DFlash fc sized wrong when num_target_layers != num_hidden_layers (#48524) Michael GoinandCodex 2026-07-21 17:42:52 -04:00
• b8fb56d970 [CI] Add gemma-4-E4B-it-assistant to CI gsm8k for GemmaMTP (#49243) Michael Goin 2026-07-21 16:51:04 -04:00
• 96a739289e [Bugfix] fix cutalss version upgrade bug, need update MSG new commit (#49016) rongfu.lengandYongye Zhu 2026-07-22 04:42:14 +08:00
• 60d443f738 [CI/Build][The Rock][BugFix] Use fork method in test_multiproc_executor_multi_node for py 3.14 compat and fix test_multiproc_executor_shutdown_cleanup (#48655) rasmithandAndreas Karatzas 2026-07-21 15:40:29 -05:00
• 1dca300653 [CI] Fix and wire encoder/manager cudagraph unit tests (#49339) Nick HillandClaude Opus 4.8 2026-07-21 20:41:16 +01:00
• fca252d59e [CI][Bugfix] Reduce max_model_len in OOT embedding test to fix KV-cache OOM on small GPUs (#49351) Flora Feng 2026-07-21 15:00:06 -04:00
• 33178f9006 Fix Qwen3-VL M-RoPE on the Transformers modeling backend (grids + compile) (#49292) Aritra Roy GosthipatyandHarry Mellor 2026-07-21 23:58:39 +05:30
• b2b8f679d0 [Bugfix][Spec Decode] Restrict embedding-width share guard to EAGLE drafts (#47953) Evan TakahashiandMichael Goin 2026-07-21 10:53:37 -07:00
• de6ec294ef [Bugfix] Fix DSA crash under breakable piecewise cudagraphs (#49302) Nick HillandClaude Opus 4.8 2026-07-21 18:28:13 +01:00
• 61e10f0116 [ROCm][CI] Fix AITER MLA fp8 decode metadata regression test (#48845) stefankoncarevic 2026-07-21 19:19:45 +02:00
• 6e96891ba0 [ROCm] Bump AITER to v0.1.16.post5 (#48683) 3 people 2026-07-21 11:39:11 -04:00
• 47f1b47a73 Ci/add laguna xs gsm8k (#49241) Michael GoinandJoe Rowell 2026-07-21 11:30:42 -04:00
• 5aab491bc9 [CI] Wire tests/models/inkling into a B200 job (#49325) Nick HillandClaude Opus 4.8 2026-07-21 15:43:20 +01:00
• 5812e1a66b [Test] Add DeepSeek MTP parallel-load tests (#41653) Stefano CastagnettaandBenjamin Chislett 2026-07-21 16:38:20 +02:00
• 8950394e0a [Bugfix] Prefix-cache metrics double-counted when a KV connector defers requests (#48860) 5 people 2026-07-21 07:02:03 -07:00
• 7bb49be4d1 [Bugfix] Handle MLA fallback during FA4 JIT warmup (#49306) Roberto L. Castro 2026-07-21 15:58:05 +02:00
• c67650f04b [XPU][DeepSeekV4]Add DeepSeek-V4 fuse_index_q SYCL kernel path (#45991) Avery MiaoandKunshang Ji 2026-07-21 21:22:34 +08:00
• f890e1dbe2 [BugFix] Set graph_pool_id before FULL CUDA graph capture in ModelRunner V2 (#48843) Ilya MarkovandMarkov Ilya 2026-07-21 14:59:26 +02:00
• 040cbf95cc [Misc] Use VLLMValidationError in chat completion tool and batch validators (#49214) Umut Polat 2026-07-21 14:38:20 +03:00