Commit Graph - vLLM - iHore Software Repository

iHore/vLLM

Fork 0

65ed0a2642 [Bugfix][GLM-5.3-Flash] Take video placeholder timestamps from the pixel path's frame sampler (#55647)
burak NNNtrance Isotr0py
2026-09-29 09:45:29 +03:00
ae0350c0de [Spec Decode][MRV2] RecoverSSM for Qwen GDN and PLE short conv (PR #58863) MOVZXandHinata 2026-09-29 13:39:09 +07:00
63b9931b73 [ROCm][DSv4.1] Paged MXFP4 sparse indexer on aiter's MQA-logits kernel (#58671)
Mehmet Cagri Claude Opus 5.5 Shanshan Shen
2026-09-29 01:37:57 -05:00
a1d3c2cf1e [Spec Decode] Fused draft metadata updates for the QSA side cache (PR #58449) MOVZXandHinata 2026-09-29 13:36:09 +07:00
cb275fdf20 [Qwen4Exp] Fix MTP buffer placement, config parsing, Triton warmup (PR #56742) MOVZXandHinata 2026-09-29 13:36:05 +07:00
e05095c9e1 [Rust Frontend] Feed parser tests and benchmarks attributed input (#59048) Bugen ZhaoandClaude Opus 5.5 2026-09-29 14:23:54 +08:00
35d6fb3187 [Perf][Spec Decode] Enable fused multi-step draft decode for FlashInfer trtllm-gen (#58371) Bowen WangandClaude Opus 5.5 2026-09-29 14:06:47 +08:00
a63dca4afb Merge remote-tracking branch 'upstream/main' into patches MOVZX 2026-09-29 12:54:18 +07:00
28f6739576 [Rust Frontend] Relax schema-aware tool argument conversion (#59004) Bugen ZhaoandClaude Opus 5.5 2026-09-29 13:40:41 +08:00
ec5e0c352f [Bugfix][Engram] Keep THP tables private when resolving shared memory (#59068) Ren YuzhouandClaude 2026-09-29 13:19:49 +08:00
17cc3cd80e [Model] Engine based plamo3 parser (#56575) Shinichi Hemmi 2026-09-29 14:17:42 +09:00
36768d1bfd [Bugfix][Frontend] Preserve caller parameters in offline pooling (#58842) shaohuaxiandCodex 2026-09-28 22:02:59 -07:00
1fde8ff7da [Bugfix] Detect the CUDA toolkit the way FlashInfer does in has_flashinfer() (#58360) Bowen WangandClaude Opus 5.5 2026-09-29 12:14:49 +08:00
13c8589b86 fix: return loaded parameters in Aria load_weights (#57476) debanshd 2026-09-28 21:05:07 -07:00
295a0b8d64 [Bugfix][Multimodal] Fix flat/scoped mm_processor_kwargs merge and resolution (#56372) Daniel M. García-Ocaña Hernández 2026-09-29 05:58:39 +02:00
491f44adfa [Model] Extend device-side mm normalization to Qwen3VL/Qwen3.5/Qwen4Next (#51289) cjackalandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-09-29 12:44:54 +09:00
07f1ba2ec6 [Multimodal] Remove processsor fallback for fused input norm params resolve (#59073) Isotr0pyandcjackal 2026-09-29 11:33:43 +08:00
aedaba8664 [CI/Build] Skip the snapshot runtime on CUDA 12.x images (#59118) Kevin H. LuuandClaude Opus 5.5 2026-09-28 20:17:59 -07:00
4e59d53ab5 [MRV2] Support stock torch.compile mode (#59079)
Cheese Cake OpenAI Codex Nick Hill Wentao Ye
2026-09-29 08:47:48 +05:30
4406874e7c [Bugfix][FT] Pass stateless process-group timeouts explicitly (#58611) Tyler Michael SmithandCodex 2026-09-28 23:14:02 -04:00
cf971b419a [Bugfix] Revert "[ROCm][Perf] Replace torch.topk in DSA candidate block selection" (#59125) Fangzhou AiandCursor 2026-09-28 22:11:48 -05:00
c6843d397c [Bugfix][Frontend] Validate reused prompt token ids; render messages for media and echo (#55771)
Shijie Lyu Shijie Lyu Claude
2026-09-28 19:36:22 -07:00
0af34418e9 [ROCm][DSv4.1][Perf] Run the delayed mHC seams through aiter's fused Triton kernel (#58655) Muhammad Ahmed 2026-09-28 22:19:10 -04:00
7230dfea50 [XPU] Preserve non-contiguous strides when pinning CPU tensors for UVA view (#54874) Chaojun Zhang 2026-09-29 09:43:07 +08:00
70ae0b7435 [ROCm][MLA] Enable sparse MLA Gluon kernel from Aiter (#53492) Mehmet Cagri 2026-09-28 20:08:18 -05:00
869278cbb0 [CI] Bound the CRCR report by build age, and stop gating it on a job (#59066) Andrey Talman 2026-09-28 20:42:32 -04:00
b39f760cc7 [torch.compile] Canonicalize functionalized split slices for fusion pa… (#57299) Laith Sakka 2026-09-28 17:21:59 -07:00
171d1b9ef6 [Elastic EP] Support Model Runner V2 (#53934) Almog Tavor 2026-09-29 02:20:40 +02:00
c68eb98b69 [Bugfix][Frontend] Keep in-flight requests on the same DP engine (#59017) Benedikt Falk 2026-09-29 01:52:29 +02:00
75fad5bbef [Bugfix][ROCm] Drop -1 sentinels when building the ragged sparse-MLA indices (#58058) Mehmet CagriandClaude Opus 5.5 2026-09-28 18:40:11 -05:00
9b29370524 [Bugfix] Fix moe_wna16 w13 zero-point shard split for 8-bit asym GPTQ MoE (#58950)
Spandan Tiwari Claude Andreas Karatzas
2026-09-28 16:38:28 -07:00
8a497d7557 [CI/Build][BugFix][The Rock] Make supports_mm_prefix return False for ROCm attn and unified attn since Prefix-LM not implemented (#52395)
rasmithmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>Andreas Karatzas
2026-09-28 18:30:38 -05:00
e03875f4d2 [Minimax M3] Enable fp8 indexer cache on triton indexer for non-SM100 architectures. (#59081)
rmhaskar Jarrel Seah Tyler Michael Smith OpenAI Codex
2026-09-28 16:21:13 -07:00
6a6c2952fa [XPU][CI] Make test_mamba_prefix_cache block-size agnostic (#58269) Fanli LinandKunshang Ji 2026-09-29 07:12:43 +08:00
3bd7302396 [XPU] skip test_online_quantization_loads_real_weights (#58987) Yan Ma 2026-09-29 07:10:52 +08:00
754aa29c00 [XPU] use rms_norm xpu kernel for context-key normalization (#58896) Yan MaandKunshang Ji 2026-09-29 07:09:58 +08:00
32cc3f1ea8 [CI] Drop test-group rules that no longer match the tree (#58895) Abhinandan 2026-09-29 03:16:21 +05:30
3d5f4d4cd5 [Perf][Spec Decode] Avoid triton recompiles in the acceptance estimator (#57107) Giancarlo DelfinandWoosuk Kwon 2026-09-28 13:46:25 -07:00
d28795f1a7 [Bugfix][Scheduler] Refresh max tokens for streaming continuations (#57676) 0xsenseiandOpenAI Codex 2026-09-29 02:01:57 +05:30
beef814e1d [ROCm][CI] Add missing test coverage for upstream parity (#50519)
Andreas Karatzas Stefan Koncarevic OpenAI Codex OpenAI Codex
2026-09-28 15:12:07 -05:00
fedbc3b564 [Bugfix][MRV2][Spec Decode] Reject draft slots that were never proposed (#58784)
qizixi Claude Nick Hill
2026-09-28 13:07:15 -07:00
42dbe4578c [Core] Rework scheduler skipped_waiting queue (#58947) Nick HillandRobert Shaw 2026-09-28 13:03:16 -07:00
2619b6b72b [Mypy] Fix mypy typing for Voxtral and vision models (#58251) Ashraf BhuiyanandWentao Ye 2026-09-28 15:52:50 -04:00
2840ca7a33 [Bugfix][Scheduler] Preserve logprobs across streaming continuations (#57447) 0xsensei 2026-09-29 01:20:27 +05:30
e7c903609f [Perf][MRV2] Allow FULL decode graphs for one-token prompt tails (#58400)
Nick Hill Nicolò Lucchesi Claude Opus 5.5mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>Yongye Zhu
2026-09-28 12:48:52 -07:00
ef42093a36 [Model Runner V2][Spec Decode] Support spec decode with draft model (#43091) Xiaoshuang Wangandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-09-29 03:38:05 +08:00
2f4a975901 [Bugfix] Fix resumable request + async scheduling handoff race (#58259) yzong-rh 2026-09-28 15:26:57 -04:00
764413559a [Bugfix] Don't sync-police or retry FlashInfer all-reduce workspace creation in eager mode (#58498)
Kevin H. Luu Claude Opus 5.5 Lucas Wilkinson Nick Hill
2026-09-28 12:14:21 -07:00
25624f65cf [ROCm][MLA] Add an AITER ASM round-robin decode route for DCP multi-token verify (#56861)
xiaohuguo2023 Claude Opus 5 seungrokj Cursor
2026-09-28 19:52:21 +01:00
5ba2e3090b [ROCm][CI] Increase timeout for Entrypoints Unit (#59076) djramicandAndreas Karatzas 2026-09-28 19:56:10 +02:00
7a160c6fe5 [ROCm][CI] Pin OpenTelemetry to LMCache's cap in the ROCm images (#59051) Rohan PotdarandClaude Opus 5.5 2026-09-28 12:33:09 -05:00
6c300ddb9e [CI][Bugfix] Relax packed_qk_rope_ correctness test to one ULP (#59008) stefankoncarevicandAndreas Karatzas 2026-09-28 19:29:59 +02:00
c01368fd64 [Quant] Use canonical N-first weight format for CT WNA16 MoE (#52798) $HDCharles <{"message":"Not Found","documentation_url":"https://docs.github.com/rest/users/emails#list-email-addresses-for-the-authenticated-user","status":"404"}>$
HDCharles Claude Opus 4.6 $HDCharles <{"message":"Not Found","documentation_url":"https://docs.github.com/rest/users/emails#list-email-addresses-for-the-authenticated-user","status":"404"}>$ HDCharles <{"message":"Not Found","documentation_url":"https://docs.github.com/rest/users/emails#list-email-addresses-for-the-authenticated-user","status":"404"}>OpenAI Codex OpenAI Codex
2026-09-28 13:22:44 -04:00
6d3ea3c2c9 [KVConnector][NIXL] Support packed MLA KV layouts in pipeline-parallel push prefill (#50499) qizixi 2026-09-28 09:50:46 -07:00
94d1462924 [Bugfix][Kimi-K3] Refresh DSpark context KV cache pointers after the KV cache is re-bound (#58814) Oxana KorzhandClaude Opus 5.5 2026-09-28 10:05:10 -06:00
b0d2b854f6 [ROCm]Keep LMCache OpenTelemetry on the image's 1.40 stack (#59056) Micah Williamson 2026-09-28 10:49:48 -05:00
6b24bd8e75 [ROCm][Perf] Replace torch.topk in DSA candidate block selection (#58208)
Fangzhou Ai TJian Cursor Agent
2026-09-28 10:42:42 -05:00
953f90d25f [LoRA] Support variable num_labels for sequence classification (#57766) linitra24andJee Jee Li 2026-09-28 22:58:35 +08:00
b721a4c709 [Feature] Add fixed-token prefill scoring (#54335)
aoshen02 aoshen02 Claude Opus 5
2026-09-28 22:50:09 +08:00
a7ff44355c [ROCm][Perf] Kimi-K3 Enable sharded latent MoE up-projection under EP (#54956) xaguilar-amdandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-09-28 16:31:29 +02:00
cd63bf9996 [Bugfix] Use the correct repository revision for secondary artifact loaders (#57461)
Clinton Thomas Lucas Bourtoulemergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-09-28 10:25:40 -04:00
d7f5722d7b [Bugfix][Logging] Preserve application log record factories (#58747) Misha GoinandOpenAI Codex 2026-09-28 09:20:16 -04:00
522101f6ed [Multimodal] Avoid extra d2d for encoder cudagraph with fused input norm (#56711) Isotr0py 2026-09-28 21:04:01 +08:00
7ba3df63cb [ROCm][Refactor] Move DeepSeek-V4/V4.1 multi-stream overlap gate to ROCm platform (#58983) Shanshan Shen 2026-09-28 20:46:28 +08:00
20b52e9f5b [Perf][Qwen3.8] Reduce PLE metadata construction overhead (#58114) Jiangyun ZhuandOpenAI Codex 2026-09-28 20:46:22 +08:00
f6877f563c [Bugfix][Spec Decode] Implement get_top_tokens() on the ROCm DeepSeek V4 MTP drafter (#57568) BaoYunkai 2026-09-28 20:16:38 +08:00
8cc9aa5ad3 [Bugfix][Model] Gemma4: register aliased embedding scalars as buffers (#54213) Yannick Schnider 2026-09-28 12:56:31 +02:00
d6cce94fd4 [Perf][Qwen4Exp] Fuse HC down projection and SiLU on NVIDIA (#58957) Thien Tran 2026-09-28 18:51:38 +08:00
252279314b Revert "[CI] Shard (H200 MIG 35GB / MI355 DPX) Entrypoints Integration (Pooling) into named jobs (#58652)" (#59011) Kevin H. LuuandClaude Opus 5.5 2026-09-28 03:51:17 -07:00
ccfd1cea75 [CPU] Include vLLM Recipes tooling in release image to deploy models using vLLM Recipes (#58796) Louie Tsai 2026-09-28 01:46:53 -07:00
a2be4d3cc7 [Bugfix][Qwen4Exp] Release the profiling KV cache held by QSA key views (#58961)
Gabriel Wu Claude Opus 5.5mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-09-28 16:11:46 +08:00
66d2d4b2e7 [CI] Split Dynamic Shapes out of (H200 MIG 35GB) PyTorch Compilation + (MI300) mirror (#58451) Thang NguyenandClaude Opus 5.5 2026-09-28 03:32:11 -04:00
413ae4b167 [CI] Shard (H200 MIG 35GB / MI355 DPX) Entrypoints Integration (Pooling) into named jobs (#58652)
Thang Nguyen Claude Opus 5.5 khluu
2026-09-28 03:21:08 -04:00
004e37ece2 [Misc] Name each backend and its kernel block sizes in block-size errors (#58557)
Eugenio "Jay" Zuccarelli Claude Opus 5.5mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-09-28 02:33:18 -04:00
4694145045 [Skills] Update kernel-microbenchmark to include ROCm (#58646) Thien TranandKimi 2026-09-28 14:18:38 +08:00
fd2f3fcfa5 [CPU] Add video inferencing via torchcodec on s390x (#58693) Rehan KhanandLi, Jiang <jiang1.li@intel.com> 2026-09-28 11:46:59 +05:30
1668e5976d [CPU] Use accelerator memory API in DiffusionGemma (#58964) zhao, zhenhuiandLi, Jiang <jiang1.li@intel.com> 2026-09-28 13:31:42 +08:00
2407f405b5 [CI] Allowlist-shrink batch 1: wire 17 root-level tests + drop 5 stale watermarking entries into misc.yaml (#58055) Turner JabbourandClaude Sonnet 5 2026-09-27 23:23:46 -06:00
76af434852 Merge remote-tracking branch 'upstream/main' into patches MOVZX 2026-09-28 12:06:52 +07:00
af7f9488c2 [Benchmark] Add Responses API backend to vllm bench serve (#54628)
Harshil Claude Chauncey
2026-09-27 22:04:11 -07:00
0da126676e [Bugfix] Fix the two multimodal root tests that fail on main (OpenPangu-VL embed merge, MiMo sink test fixture) (#58900) Kevin H. LuuandClaude Opus 5.5 2026-09-27 21:52:51 -07:00
dd3bc9c72f [Perf][MRV2] Reuse Mamba/GDN metadata across KV cache groups (#58762)
Luca Motz Xin Yang Claude Opus 5.5 Codex Lucas Wilkinson
2026-09-28 06:36:58 +02:00
55de40a2fc [Minimax-M3] Add Encoder CUDA graph support (#58673) Thien TranandCodex 2026-09-28 12:17:36 +08:00
6f65fde419 [Bugfix][Frontend] Apply Harmony adjust_request in batched chat completions (#58958) Flora Feng 2026-09-28 00:12:04 -04:00
4a190dc518 [Model] Enable LoRA support for RobertaForSequenceClassification (#58884) jz-yoloandjz-yolo 2026-09-27 20:50:12 -07:00
77fbd9e225 [Test][ROCm] Stabilize the mixed OLMoE LoRA test (#58945) Andreas Karatzas 2026-09-27 22:47:41 -05:00
5eaa50f151 [CPU] Build CPU wheels on Ubuntu 22.04 with AMX-FP8 support (#58515) zhao, zhenhuiandjiang1.li 2026-09-28 11:46:02 +08:00
53d2e16a97 [Bugfix][EPD] Skip sampling for encoder-only async steps (#58490) Tianyu Guo 2026-09-28 11:35:09 +08:00
027b6f3a22 [Bugfix][Frontend] Use a fresh parser per choice in non-streaming chat completions (#58939) Flora Feng 2026-09-27 22:01:35 -04:00
4d07e90653 [Bugfix][Frontend] Sample batched chat completions from the adjusted requests (#58929) Flora Feng 2026-09-27 21:46:55 -04:00
b9e2903b89 [Mypy] Fix mypy typing for Ultravox and Unlimited-OCR models (#58239) Ashraf Bhuiyan 2026-09-27 21:19:04 -04:00
39d49539de [KimiViT][Perf] Fuse per-layer QK RoPE into one in-place kernel (#58651) Thien Tran 2026-09-28 09:18:06 +08:00
3184226984 [Bugfix][Frontend] Count Responses reasoning tokens per tool round (#58927) Flora Feng 2026-09-27 20:34:38 -04:00
31afa4f1bb [ROCm] Bump AITER to v0.1.23 (#58867)
Fangzhou Ai Cursor Andreas Karatzas
2026-09-27 19:04:58 -05:00
73859fec58 [Frontend] Switch Python Harmony dependency to oss-harmony (#55128)
Anton Peganovmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>OpenAI Codex
2026-09-27 22:43:56 +02:00
386ac2573e [Bugfix][KV Connector] Retry Mooncake bootstrap registration on timeout (reopens #55763) (#58919) Chao Lei 2026-09-28 03:47:36 +08:00
187c81b32d [ROCm][Bugfix] Fall back to default GEMM for CPU tensors on ROCm builds (#58923) Fangzhou Ai 2026-09-27 14:47:07 -05:00
231fdb83cc [Perf][MoE] Use fused MiniMax2 routing with non-unit routed scaling (#58880) Jiangyun Zhu 2026-09-28 03:37:33 +08:00
e7900156e1 [Fast Start] Add /health endpoint for the weight cache daemon (#58552) Xun SunandIsotr0py 2026-09-28 01:48:54 +08:00
2b9b55c7f1 [Bugfix] Support repsonse_format + tool_choice=auto (#56086)
Artem Perevedentsev Yufeng He pablopupo hubunt Claude Sonnet 5
2026-09-27 20:17:34 +03:00

... 11 12 13 14 15 ...

Commit Graph Select branches Hide Pull Requests main patches sm86 Mono Color

Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86