Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• eeaf23107f [ROCm] Add tuned selective_state_update float32 config for AMD Instinct MI300X (#47947) vanshbhatia-amdandvanshbhatia-amd 2026-07-08 07:09:02 -04:00
• 285c08c036 [Model] Support MOSS-Transcribe-Diarize (#47729) Canlin Guo 2026-07-08 19:05:45 +08:00
• 1f4ad059d1 [ROCm] Add tuned selective_state_update float16 config for AMD Instinct MI300X (#47945) vanshbhatia-amdandvanshbhatia-amd 2026-07-08 07:03:58 -04:00
• 04a703e397 [Frontend] Support bad_words in the /v1/completions endpoint (#46793) sungbin1015andClaude 2026-07-08 18:51:17 +09:00
• bd3bb4eb26 [Misc][Docs] Add human-readable integer support for more cli-args (#47608) Nicolò Lucchesi 2026-07-08 11:43:34 +02:00
• 440002552e [XPU] [Fusion passes] Disable fuse_rope_kvcache_cat_mla & qk_norm_rope_ fusion on XPU (#47962) Chaojun Zhang 2026-07-08 17:23:06 +08:00
• 99a85617bf [Test] Skip DeepEP MoE layer tests without P2P access (#47946) Tyler Michael SmithandClaude 2026-07-08 04:46:03 -04:00
• 7c67da967f Remove unused _get_kv_cache_config_deepseek_v4 alias (#47969) Nicolò Lucchesi 2026-07-08 10:18:03 +02:00
• d79855eaac [Docs] kv_sharing_fast_prefill correction (#47044) Nicolò Lucchesi 2026-07-08 10:17:46 +02:00
• 51e5372f3d [Model][HunyuanVL] Use native transformers processor and adapt to transformers 5.13 (#47872) manayangandmanayang 2026-07-08 15:58:23 +08:00
• 7cc2e8e74f fix: hash speculative draft model config (#47911) Ace Eldeib 2026-07-08 03:36:30 -04:00
• 2c64b4c1cc [ROCm] fixed aiter master flag and expert parallelism compatibility on minimax-m3-mxfp8 (#47158) Hongxia Yang 2026-07-08 03:26:17 -04:00
• d35eba302f [Bugfix] Avoid leaking Pydantic repr in tool_choice error message (#47028) Muhammad FawazandMahad Durrani 2026-07-08 00:00:59 -07:00
• c0e8e1f12a [Bugfix] Register VLLM_BUILD_* and VLLM_IMAGE_TAG provenance env vars (#45313) Nicklas Frahm 2026-07-08 08:21:12 +02:00
• 5d5fab0061 [Bugfix][Frontend] Fix http_requests_total metric recording some 4xx errors as 5xx (#44303) Zach Zhu 2026-07-08 13:33:21 +08:00
• 2afa3f7e95 [Perf] Minimax M3 - Support cross-layer allreduce-norm fusion (#47631) Wei ZhaoandYongye Zhu 2026-07-08 00:16:32 -04:00
• 80eb01e93d [Bugfix] DSV4 TP16 garbage output (#47493) Jeff (Junze) MaandLucas Wilkinson 2026-07-07 21:04:33 -07:00
• d9e57ea82e [ROCm][Perf] MXFP8 dense-linear + grouped-MoE GEMM optimizations for MiniMax-M3 (#46117) Ethan Yangandamd-ethany 2026-07-08 12:03:34 +08:00
• 9021589498 [Minimax-M3] Using tok_sparse_select from MSA instead of triton kernels (#47502) Yongye ZhuandClaude Opus 4.8 2026-07-07 21:01:12 -07:00
• 0303f37a54 [Bugfix][Pooling] Align CrossEncoder token type ids after truncation (#47772) Ting SUN 2026-07-08 10:59:22 +07:00
• dd127d82ed [Core][Engine] only materialize tokens when thinking budget is in req (#47053) Walter Beller-Morales 2026-07-07 23:02:38 -04:00
• 0ca6eee743 [Core] Pass request context to CPU offload cache policy touch (#47744) Posedge_LinandOr Ozeri 2026-07-08 10:56:25 +08:00
• 5e975eae1a [Bugfix] Avoid blocking model launching when no system ffmpeg available for TorchCodec (#47888) Isotr0py 2026-07-08 10:52:25 +08:00
• f7fc0ca993 [Frontend] Add endpoint plugins framework (#47454) Martin Hickey 2026-07-08 03:00:41 +01:00
• f7efab58ec [CPU][Bugfix] Fix flaky ShortConv prefill test on ARM (uninitialized weights) (#47848) Rahul Vishwakarma 2026-07-08 06:50:09 +05:30
• e97c3cb303 [Core] Persist and reuse the memory-profiling result across boots (opt-in) (#47388) 3 people 2026-07-07 18:53:02 -06:00
• 4aceabf8c1 [ROCm][Bugfix] Key sparse-MLA persistent metadata on per-request context lengths (#47766) Rohan PotdarandClaude Opus 4.8 2026-07-07 19:22:34 -05:00
• 6e35c5e5af [ROCm][CI] Minimize comment in RocmAttention q_scale check (#47731) stefankoncarevic 2026-07-08 02:16:08 +02:00
• aad0fb741b [CI/Build] Accept ready-run-all-tests label in pre-commit gate (#47897) 3 people 2026-07-07 16:18:59 -07:00
• 7d2ce5750e [Bugfix] Patch Hopper MXFP4 OOB scales reads leading to NaN (#47910) yzong-rh 2026-07-07 18:51:22 -04:00
• 675f4295cd fix(security): bound completion prompt list to prevent unbounded engine fan-out (#47845) Juan Pérez de Algaba 2026-07-08 00:48:20 +02:00
• d99adcebdc [BugFix] Fix ModelOpt quantization inference for fused siblings (#47445) Jason Li 2026-07-07 18:19:42 -04:00
• c8c2f838e7 Add tuned selective_state_update config for AMD Instinct MI355 (#47767) vanshbhatia-amdandvanshbhatia-amd 2026-07-07 18:19:08 -04:00
• dd0d74cd92 [Doc] Surface the --kv-cache-memory suggestion at INFO and document fast-startup knobs (#47374) 3 people 2026-07-07 16:05:07 -06:00
• 55da232db6 [Bugfix] Pad Mamba page size instead of scaling block_size in unify_kv_cache_spec_page_size (#45207) Sahil KadadekarandClaude Fable 5 2026-07-07 18:01:34 -04:00
• 3f99883d97 [CI Bug Fix] Temp fix for v3.2 accuracy (#47902) Wentao Ye 2026-07-07 16:36:03 -04:00
• 47c40bfe8a [Doc] Fix manylinux tag in installation guide (#47913) Nick Cao 2026-07-07 16:34:05 -04:00
• 3dd910da42 [Bugfix] Allow non-contiguous query in FlashInfer FP8 query quantization (#47908) Matthew BonanniandClaude Opus 4.8 2026-07-07 16:11:34 -04:00
• 7bd154375d [Bugfix] Fix mamba+dflash for MRV2 (#47698) Benjamin Chislett 2026-07-08 04:59:13 +09:00
• 2f3f441f84 fix: include topic frame in KV events replay response (#45177) Rishabh Saini 2026-07-07 14:48:23 -04:00
• d6875196ad [Bugfix] Exclude kv_cache_memory_bytes from CacheConfig.compute_hash (#47356) Nils MattesonandNils Matteson 2026-07-07 11:46:51 -06:00
• abe41f28de Upgrade tpu-inference to v0.24.0 (#47835) Sting Lin 2026-07-08 01:15:32 +08:00
• c3284c31f5 [Perf][3/N] Expand Triton kernel warmup coverage, Qwen (#47546) Roberto L. Castro 2026-07-07 19:06:59 +02:00
• c74e751824 [Doc] Fix grammatically incorrect error message in gpu_worker and xpu_worker (#36715) Robin 2026-07-07 18:03:06 +01:00
• b93cbd7416 [XPU] Fix topk_sigmoid arg mismatch on XPU (#47858) liuzhenwei 2026-07-08 00:53:18 +08:00
• bdc6f3bfa1 [Bug] Fix tmp directory for lm_eval (#47755) Wentao Yeandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-07 12:40:12 -04:00
• 392d1b4d2e [BugFix][LoRA] Refresh punica metadata when LoRA slots are reassigned under an unchanged mapping (#47725) 3 people 2026-07-07 08:53:55 -07:00
• 21b396abe1 AGENTS MD: Add suggestion on how to incorporate tests (#47784) 3 people 2026-07-07 08:16:08 -07:00
• bdaf27519f [XPU] Fix Event init failure w/ blocking (#47868) liuzhenwei 2026-07-07 22:54:03 +08:00
• beb4327c46 Enable causal masking for SWA in vllm-project/speculators models (#47745) Eldar Kurtić 2026-07-07 16:24:14 +02:00
• c46ced1ee3 [kv_offload] Establish tier-owned KV event handling (#46544) 3 people 2026-07-07 06:55:20 -07:00
• 65dcde1695 [Bugfix] Fix PD disagg + MTP correctness for Qwen3.5(GDN) (#47466) Dakai AnandNick Hill 2026-07-07 21:51:22 +08:00
• 65a7b46284 [KV-Offloading] Support workload identity for objectstore secondary tier (#47063) Pierangelo Di PilatoandOr Ozeri 2026-07-07 15:30:16 +02:00
• 93e2ab7111 Disable dynamic speculative decoding when DP is enabled (#45963) 3 people 2026-07-07 08:52:41 -04:00
• 8b745527cd [Bugfix] Fix UBatchWrapper CUDA graph key to sum all ubatches, not just first two (#43161) Lanze Liu 2026-07-07 05:42:31 -07:00
• 920469974a [UX] Log worker exit code when process dies unexpectedly (#38641) Nick CaoandClaude 2026-07-07 08:36:29 -04:00
• 8b91cd5b20 [Bugfix][Core] Close underlying iterator in merge_async_iterators single-iterator fast path (#44726) Ting SUNandNick Hill 2026-07-07 19:13:41 +07:00
• dd94484577 Bump Transformers version to 5.10.4 (#41359) Harry Mellor 2026-07-07 13:13:28 +01:00
• 7ff656cc8b fix: ensure no double load of lm head in nemotron mtp (#47440) Shaun Kotek 2026-07-07 15:01:45 +03:00
• 0a2965b1b3 [BugFix] Fix ModelOpt mixed-precision quantization for sparse quantized_layers configs. (#47318) danielafrimi 2026-07-07 14:45:13 +03:00
• 0ed05b6f82 [CI] Fix Transformers modeling backend LoRA test (#47832) Harry Mellor 2026-07-07 12:40:00 +01:00
• ed051fab54 [Bugfix] Reject sampling params unsupported by diffusion models (#45418) Guan-Ming Chiu 2026-07-07 19:25:36 +08:00
• 48fcfc926c [KV Offload] Add ParentManager ABC for secondary tier callbacks (#47274) Ronen SchafferandOr Ozeri 2026-07-07 13:51:18 +03:00
• 3354dba381 [Bugfix][KV offload] Store interior chunk-boundary blocks under MTP/Eagle (#46972) 3 people 2026-07-07 13:16:52 +03:00
• cbb5f045be [ROCm][CI] Refresh ROCm base images when docker rocm_base changes (#46904) 5 people 2026-07-07 05:10:50 -05:00
• b3e85be663 fix: use configured max_logprobs instead of hardcoded 20 in derender validation (#47834) Juan Pérez de AlgabaandCursor 2026-07-07 11:42:47 +02:00
• d3e69fd671 [Perf] Use blocking CUDA events to avoid busy polling cuda driver lock (#47081) Summer Yang 2026-07-07 02:36:10 -07:00
• c85d72076a [HARDWARE][POWER] optimize math functions of VSX power (#47321) 3 people 2026-07-07 15:05:47 +05:30
• c5b66233b2 [Bugfix][Spec Decode] Skip uniform spec-decode padding for diffusion models (#47464) Kyung Sub Lee (Daniel)andClaude Fable 5 2026-07-07 04:25:12 -05:00
• 066f02ae94 [MoE] FI autotuning: max bucket = max token count [e.g. DP_size*MNBT] (#47427) Netanel Haberandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-07 12:08:36 +03:00
• 5d23ca47ab [Kernel] Applies routed_scaling_factor internally (#47408) Jee Jee Li 2026-07-07 17:00:54 +08:00
• e55cc59e52 [Rust Frontend][CI] Unblock more end-to-end test cases (#47735) Bugen Zhaoandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-07 16:27:20 +08:00
• ba50b9763f [Bugfix] Match the mapped filename in find_loaded_library (#47586) 3 people 2026-07-07 16:06:29 +08:00
• b4cfbc24d3 [Bugfix][Core] Fix host memory leak from undrained new_block_ids (#44490) 3 people 2026-07-07 14:32:55 +07:00
• 1e823dc01d [docs update] Update usage of hf cli for cache list and removal (#47830) Aritra Roy Gosthipaty 2026-07-07 12:39:18 +05:30
• 8e61b646e2 fix(security): add resource bounds validation to derender endpoints (#47260) Juan Pérez de Algabaandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-07 08:58:26 +02:00
• e040899a00 [KV Offloading] Add basic offloading metrics (#45958) 4 people 2026-07-07 01:26:28 -05:00
• dd5c299fbe [ROCm][Bugfix] Convert ModelOpt FP8 per-channel weights to e4m3fnuz on MI300/MI325 (#47201) Micah WilliamsonandAndreas Karatzas 2026-07-07 01:24:58 -05:00
• 6db31c8e76 [XPU][CI]Adjust memory request for tests in Intel GPU CI (#47758) xiangdong 2026-07-07 13:56:40 +08:00
• cbe9c40f99 [Bugfix] Forward callable hf_overrides to the draft model config (#45352) 3 people 2026-07-06 23:12:38 -05:00
• 2f71b2bd9f [ROCm] Align mixed encoder-decoder KV cache views in V2 runner (#47685) Andreas Karatzas 2026-07-06 23:09:22 -05:00
• 32ab064621 [UX] Add model_class_overrides for development and debugging (#47148) Jee Jee Liandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-07 11:43:08 +08:00
• c64c356990 [Perf] Bound DiffusionGemma sampler transient via request-tiled logits (#45672) Guan-Ming Chiu 2026-07-07 11:42:01 +08:00
• 34e6dfced8 [Rust Frontend] Stamp arrival_time at the frontend entry (#47787) Tahsin Tunan 2026-07-07 09:27:10 +06:00
• 39a1d32b59 [Rust Frontend] Avoid extra copies for multimodal tensors (#47581) Reid 2026-07-07 11:09:55 +08:00
• 700e882eab Add TorchCodec as a video decoding backend (#46609) 3 people 2026-07-07 03:58:51 +01:00
• a4f019fa25 fix(distributed): propagate distributed_timeout_seconds to NCCL device groups (#45159) jialoop-gitandClaude Opus 4.8 2026-07-07 10:52:51 +08:00
• 9dd2465896 feat(cpu): add CPU support for Mamba ShortConv (#35059) Rahul Vishwakarma 2026-07-07 08:17:08 +05:30
• a46c9329e5 [Rust Frontend] Add DeepSeek V3.2 roundtrip fixture (#47619) Reid 2026-07-07 10:47:02 +08:00
• 445321fab4 [Bugfix] [Quantization] Fix loading for CT DSV2 (#47780) Kyle Sayers 2026-07-06 22:28:00 -04:00
• 69f3150981 [XPU] Fix PP accuracy on XPU device (#47253) YiSheng5andKunshang Ji 2026-07-07 09:17:27 +08:00
• 86db6c3070 [Frontend] add per-request timing metrics field to response body of Chat/Completions APIs (#46768) 3 people 2026-07-06 19:48:29 -05:00
• 5769a7382c [ROCm][CI][Bugfix] Fix flaky parallel tool-call streaming (test assertion + Mistral/Granite parsers) (#47550) akii96andAndreas Karatzas 2026-07-07 02:06:19 +03:00
• 8484ca5d45 [ROCm][CI] Adding Rust parity (#47478) Andreas Karatzas 2026-07-06 17:05:39 -05:00
• 482e5524fe [Bugfix][ROCm] Fix memory access fault in AITER MLA backend for DPA+FP8 KV (#47276) Simon Danielssonandnnyrhila 2026-07-06 23:30:02 +02:00
• 567a78432d [Bugfix] Fix dp mtp hang (#40589) 3 people 2026-07-07 05:08:17 +08:00
• d891b9bd51 [Quantization] add humming moe backend to all dense/moe oracles (#41652) Jinzhen Linandmgoin 2026-07-07 04:36:07 +08:00
• 04adc8843b [Bugfix]Fix DeepSeek-V4 fp8_ds_mla KV cache reshape (#47716) 3 people 2026-07-07 03:56:44 +08:00
• ae098abe3f [CI] Fix some errors on main (#47726) Harry Mellor 2026-07-06 20:40:23 +01:00
• b1384f5ec6 Enable B12x backend for non-gated MoEs (like Nemotron) (#43328) Andrii SkliarandAndrii Skliar 2026-07-06 21:40:07 +02:00