81 Commits
Author SHA1 Message Date
MOVZXandHinata 5aa2ac64f8 [Mini-VLLM] Remove LoRA, plugins, speech-to-text and scale-out surfaces
Production runs one static model behind the OpenAI-compatible API. Every
adapter path and non-OpenAI HTTP surface is unreachable there.

- Delete the LoRA stack: vllm/lora, LoRARequest on the engine wire, the
  scheduler, worker, cudagraph and MoE plumbing, LoRA metrics, the lora
  CLI group, and the LoRA tests.
- Drop the plugin loader in vllm/plugins. Platform resolution is now
  CUDA-only and the general/stat-logger plugin hooks are gone.
- Delete the speech-to-text and realtime entrypoints, the scale-out
  render routes, and vllm/config/speech_to_text.py. GenerationTask is
  generate-only.
- Cut DecodeBenchConnector and the disagg-prefill doc and image. The
  Multi, Offloading and Example connectors stay; production uses the
  CPU KV offload tier.
- Delete the ROCm-only fused KDA chunk kernel (2346 lines) with its CMake
  arch gate, declarations and op registrations. The CUDA build never
  defined the guard and no Python caller existed.
- Delete the orphan Mistral and Pixtral chat fixtures and examples/.
- Fix _check_model: removing the LoRA early return made every request
  fail with "The model ... does not exist. Valid aliases: ...".

Verified on an RTX 3090: import sweep reports 1242 modules with 7 known
optional-dependency failures; a live server answers chat, completions,
responses, tokenize and detokenize with zero log errors and 404 for the
removed surfaces; collection yields 8558 tests with the 3 pre-existing
errors; ruff adds no new finding; the incremental _C build succeeds.

Co-authored-by: Hinata <hinata@idihore.id>
Signed-off-by: MOVZX <movzx@yahoo.com>
2026-10-19 21:06:19 +07:00
MOVZX 76df77c04e Merge branch 'main' into patches
Signed-off-by: MOVZX <movzx@yahoo.com>
2026-10-15 21:16:39 +07:00
FeanorandTaneem Ibrahim 5d5c3d7c43 [Bugfix] Preserve padding token types for cross encoders (#58143)
Signed-off-by: zhaoye <772971548@qq.com>
Co-authored-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
2026-10-15 03:09:24 +00:00
dingdangmaoandCodex 124b89d5b7 [Bugfix][RecoverSSM] Correct aligned state indices at exact block boundaries (#59962)
Signed-off-by: apex-mochen <2756823972@qq.com>
Co-authored-by: Codex <noreply@openai.com>
2026-10-12 02:08:14 +00:00
DimensionSTPandOpenAI Codex 10cc2f6ae2 [Bugfix] Bound GDN speculative decode query widths (#55504)
Signed-off-by: DimensionSTP <65501090+DimensionSTP@users.noreply.github.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
2026-10-10 11:00:16 +08:00
Nick Hill 7d1d8660d6 [MRV2] Miscellaneous code structure updates (#60754)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
2026-10-09 17:58:09 -07:00
MOVZX ac305c44cb Merge branch 'main' into patches
Signed-off-by: MOVZX <movzx@yahoo.com>
2026-10-09 17:41:13 +00:00
3d6853b32c [Bugfix] Seed hybrid mamba state index with mamba_block_size on prefix-cache hits (#55601)
Signed-off-by: pondzikk <pondzik@gmail.com>
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: ptorsten <1652174+ptorsten@users.noreply.github.com>
2026-10-07 17:09:50 +00:00
8a26869bca [Bugfix] Index Mamba align state copies by request slot in MRV2 (#60210)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Karl0007 <44745209+Karl0007@users.noreply.github.com>
Co-authored-by: Yongqin Wang <yongqinwang@roblox.com>
2026-10-07 02:32:46 +00:00
Juntian LiuClaude Opus 5.5mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
eb2901b735 [Perf][DSv4.1] Run decoder replay layers in CUDA graphs and trim in PIECEWISE graphs (#59532)
Signed-off-by: Juntian Liu <Juntianl777@gmail.com>
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-10-05 22:11:45 -07:00
Vadim GimpelsonCursor Agentmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
0eac152707 [Bugfix] Route every speculation-capable row through the speculative path so recurrent state stays consistent (#56531)
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-10-06 03:03:44 +04:00
MOVZXandHinata ae0350c0de [Spec Decode][MRV2] RecoverSSM for Qwen GDN and PLE short conv (PR #58863)
Intake of open upstream PR vllm-project/vllm#58863. Active only with
VLLM_USE_V2_MODEL_RUNNER=1; inert otherwise.

Speculative decoding on Gated DeltaNet snapshotted the full recurrent
state after every draft position and reserved one Mamba block per draft
position. On a 24 GB card that is reserved VRAM we never use, plus
per-step copy traffic.

RecoverSSM, the protocol Kimi-K3 KDA already uses, replaces this:
verify runs from a read-only checkpoint and stores a small per-token
replay record [correction (V), normalized key (K), gate (1)] in fp32;
commit replays only the accepted tokens and writes the recurrent state
once, plus the block-boundary state in align mode. The PLE dilated
short conv compacts its accepted window with the same plan.

Conflict in vllm/models/qwen4_exp/nvidia/mtp.py: our get_top_tokens()
work put LocalArgmaxMixin on Qwen4ExpMTP, the PR adds SupportsReplaySSM
to the same base list. Kept both.

Co-Authored-By: Hinata <hinata@idihore.id>
Signed-off-by: MOVZX <movzx@yahoo.com>
2026-09-29 13:39:09 +07:00
Nick HillNicolò LucchesiClaude Opus 5.5mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>Yongye Zhu
e7c903609f [Perf][MRV2] Allow FULL decode graphs for one-token prompt tails (#58400)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Nicolò Lucchesi <nicolo.lucchesi@mistral.ai>
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
2026-09-28 12:48:52 -07:00
5567cc1b25 [Mamba] RecoverSSM align: keep a boundary-ending state in its block
A speculative window whose accepted tokens end exactly on a recurrent block
boundary selected the following block-table column for the committed state
(commit plan) and recorded it as the running column (postprocess). That
column can still be unallocated, so the next step reads an uninitialized
page. Select the block that holds the last committed token, matching the
(n - 1) // block convention used elsewhere in align mode. Applies to
Kimi-K3 KDA and to the Qwen GDN/PLE RecoverSSM paths that share the plan.

Fix and test from local-inference-lab/vllm a02f585d (Martin Vit, root cause
reported by astra); the GDN/PLE align tests now expect the same convention.

Signed-off-by: Jürgen Schmied <juergenschmied70@gmail.com>
Co-authored-by: Martin Vit <martin@voipmonitor.org>
Co-authored-by: Claude <noreply@anthropic.com>
2026-09-27 11:12:40 +02:00
Jürgen SchmiedandClaude a80e41d690 [Mamba] Share RecoverSSM commit kernels and helpers across models
RecoverSSM's commit-plan and conv-window compaction kernels are model independent.
Move them from models/kimi_k3 into model_executor/layers/mamba/ops/recoverssm.py
(Kimi-K3 keeps its private names as import aliases; kernel source unchanged), add
mamba/recoverssm_utils.py (activation predicate, argument validation) and the
generic spec-row/request mapping helpers to v1/attention/backends/recoverssm_metadata.py.

CacheConfig.use_kda_recoverssm becomes use_recoverssm: the switch now selects RecoverSSM
for any model whose recurrent layers implement it, not only KDA.

No functional change for Kimi-K3.

Signed-off-by: Jürgen Schmied <juergenschmied70@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
2026-09-27 00:07:04 +02:00
Nick HillandClaude Opus 5.5 7dbd0a8d22 [Bugfix][MRV2] Treat padded prompt tails as spec-decode rows for hybrid models (#58434)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-24 18:29:53 -07:00
Yejing LaiandKunshang Ji 14f98cbe5d [Bugfix][XPU] store the pointer raw bit pattern instead of its numeric value (#54514)
Signed-off-by: Lai, Yejing <yejing.lai@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
2026-09-24 02:05:26 -07:00
Nick Hill afec265bca [MRV2] Miscellaneous code cleanup (#57980) 2026-09-23 01:51:52 -07:00
Albert Chengmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>Benjamin Chislett
88aa0d287d [Spec decode] Support variable-length decode for Kimi-K3 adaptive ver (#52988)
Signed-off-by: Albert Cheng <albecheng@nvidia.com>
Signed-off-by: Albert Cheng (Engrg-Hardware 1) <albecheng@login-bia01.bia.clusters.nvidia.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Co-authored-by: Benjamin Chislett <bchislett@nvidia.com>
2026-09-22 20:43:32 -07:00
hclandmgoin 71a2392845 [Bugfix][V1] Read ModelState max_model_len from model config (#58149)
Signed-off-by: Chenglun Hu <chenglunhu@gmail.com>
Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
2026-09-22 19:45:04 +00:00
1dc2d854c1 [Feat][Model] Support encoder-side SWA-bounded replay for DeepSeek-V4.1-Flash (#56227)
Signed-off-by: Yifan Qiao <yifanqiao@inferact.ai>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Co-authored-by: Woosuk Kwon <woosuk@inferact.ai>
2026-09-18 15:01:24 -07:00
Harry MellorandClaude Opus 5 651a88c09a [CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-16 10:57:24 -07:00
Harry MellorandClaude Opus 5 8c1557a79c Add pydocstyle to the ruff rules (#52136)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Harry Mellor <hej.mellor@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-16 16:44:29 +00:00
ebe1dec2da [PCP][DCP] Enable PCP+DCP on sparse-MLA models (#56157)
Signed-off-by: patryk.saffer <patryk.saffer@blc-login-1.blc1.mistralcompute.ai>
Signed-off-by: patryk.saffer <patryk.saffer@blc-login-0.blc1.mistralcompute.ai>
Signed-off-by: PatrykSaffer <patryk.saffer@mistral.ai>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: patryk.saffer <patryk.saffer@blc-login-1.blc1.mistralcompute.ai>
Co-authored-by: patryk.saffer <patryk.saffer@blc-login-0.blc1.mistralcompute.ai>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Patryk Saffer <PatrykSaffer@users.noreply.github.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-12 23:24:44 +00:00
Nick Hill e6cb56337b [Core] MRV2 support for fast-prefill (#56145)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
2026-09-10 10:36:04 -07:00
Harry MellorandClaude Opus 5 719284fe15 [Core][Model] Unify XD-RoPE into M-RoPE and derive the channel count (#56078)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-09 09:23:30 -07:00
3f41d102c5 [1/2][Model Runner V2] DBO support, eager mode (#50945)
Signed-off-by: specture724 <specture724@gmail.com>
Signed-off-by: Anjie Hou <149605198+specture724@users.noreply.github.com>
Co-authored-by: jiangkuaixue123 <jiangxiaozhou111@163.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
2026-09-04 14:17:39 -07:00
Jiangyun Zhu 98ed0856f3 [Model] add GLM-5.3-Flash support (#53906) 2026-09-03 09:40:35 -07:00
+5 e126687a9a [Model] Support Qwen3.8-Flash-Next (#53896)
Signed-off-by: huanghaoyan.hhy <huanghaoyan.hhy@alibaba-inc.com>
Signed-off-by: andyluo7 <andy.luo@amd.com>
Signed-off-by: lishunyang12 <lishunyang03@gmail.com>
Signed-off-by: Harry Huang <huanghaoyan.hhy@alibaba-inc.com>
Signed-off-by: Thien Tran <gau.nernst@yahoo.com.sg>
Signed-off-by: Isotr0py <Isotr0py@outlook.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Yifan Qiao <yifanqiao@inferact.ai>
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: KungYork <firelikest@163.com>
Co-authored-by: Thien Tran <gau.nernst@yahoo.com.sg>
Co-authored-by: Yifan Qiao <yifanqiao@inferact.ai>
Co-authored-by: andyluo7 <43718156+andyluo7@users.noreply.github.com>
Co-authored-by: Isotr0py <mozf@inferact.ai>
Co-authored-by: andyluo7 <andy.luo@amd.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
Co-authored-by: lishunyang12 <lishunyang03@gmail.com>
Co-authored-by: Vadim Gimpelson <156319763+vadiklyutiy@users.noreply.github.com>
Co-authored-by: Isotr0py <Isotr0py@outlook.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
2026-08-31 05:57:56 +00:00
Wentao Ye 41729fc53b [K3 Perf] Optimize k3 mamba metadata preparation, 6.6~7.6x kernel performance improvement (#52388)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
2026-08-25 11:00:28 -04:00
7ca336929c [Model] Remove ten deprecated model architectures (#53608)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Xianbao QIAN <xianbao.qian@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-24 17:06:00 -07:00
JensTonoken3mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
da329cc303 [Bugfix] Fix speculative decoding for short_conv (LFM2) models (#50272)
Signed-off-by: jens.luecke <jens.luecke@liquid.ai>
Co-authored-by: Tonoken3 <sakamakismile@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-08-22 13:34:22 +00:00
Canlin GuoandNick Hill 455edc022b [ModelRunnerV2] Support prompt embeds (#42963)
Signed-off-by: gcanlin <canlinguosdu@gmail.com>
Signed-off-by: Canlin Guo <961750412@qq.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
2026-08-17 21:58:50 +00:00
Jiangyun ZhuOpenAI CodexBenjamin Chislettmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
70afdedc10 [K3] support recoverssm for K3 (#51855)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Benjamin Chislett <bchislett@nvidia.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-08-17 19:34:14 +08:00
Taneem Ibrahim fdab2b10bc [ModelRunner v2] Support Transformers pooling model (#52425)
Signed-off-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
2026-08-16 15:25:49 -04:00
Nick Hill d6f17f3d53 [MRV2] Support attention-free models (#52374)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
2026-08-14 23:54:50 +00:00
Isotr0pyandOpenAI Codex 66728feb1f [MRV2][Multimodal] Enable encoder cuda graph for model runner v2 (#49852)
Signed-off-by: Isotr0py <Isotr0py@outlook.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
2026-08-14 09:33:21 +00:00
Lucas WilkinsonOpenAI CodexClaude Opus 5Benjamin Chislettmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>Nick HillOpenAI Codex
7f7a32cfec [Spec Decode] DSpark confidence-scheduled verification (#47808)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Signed-off-by: Benjamin Chislett <chislett.ben@gmail.com>
Signed-off-by: Lucas Wilkinson <wilkinson.lucas@gmail.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Benjamin Chislett <chislett.ben@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
2026-08-12 12:27:48 -07:00
Guan-Ming Chiuandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 457a5f34db [Bugfix][MRV2] Support encoder timing stats in model runner V2 (#50020)
Signed-off-by: Guan-Ming (Wesley) Chiu <105915352+guan404ming@users.noreply.github.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-08-12 00:00:49 +08:00
Tianyu Guo dd856e48bb [Bugfix][EPD][Model Runner V2] Skip gather mm embeddings for encoder only instance (#51222)
Signed-off-by: Tianyu Guo <guoty@inferact.ai>
2026-08-06 20:40:47 -07:00
Varun Vinayak ShenoyandNick Hill e578de311c [ModelRunnerV2] Fix scalar Mamba state update with int32 mappings (#50327)
Signed-off-by: Varun Shenoy <varun.vinayak.shenoy@oracle.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
2026-08-03 22:05:12 +00:00
Martin Hickey b2fb83e7ff [Attention]: Use KVCacheSpec for AttentionMetadataBuilder type hints (#50148)
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
2026-07-31 11:20:18 +00:00
Taneem Ibrahim 43eaefba5a [ModelRunner V2] Enable sequence pooling for embedding and classification models (#48791)
Signed-off-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
2026-07-29 06:30:45 -07:00
f51193b9ae [Kernel][Mamba] Fused-kernel support for align-mode DS-conv state migration with num_accepted_tokens > 1 (#49291)
Signed-off-by: Sungsoo Ha <sungsooh@nvidia.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Jiangyun Zhu <riverclouds.zhu@qq.com>
2026-07-29 18:54:52 +08:00
Li, Jiang 90245f4190 [Bugfix] Fix multi-modal support on CPU MRV2 (#50073)
Signed-off-by: jiang1.li <jiang1.li@intel.com>
2026-07-28 06:09:48 +00:00
Nick HillandBenjamin Chislett 439f336212 [Core] Fix gpu<->cpu syncs in MRV2 mamba_hybrid.py (#49736)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Benjamin Chislett <bchislett@nvidia.com>
2026-07-27 02:41:27 +00:00
Nick HillandTaneem Ibrahim fdaa0d9e59 [ModelRunner V2] Support encoder-only attention (#49331)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Signed-off-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
Co-authored-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
2026-07-27 00:38:44 +00:00
Nick HillandClaude Opus 4.8 de6ec294ef [Bugfix] Fix DSA crash under breakable piecewise cudagraphs (#49302)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-21 18:28:13 +01:00
b6ff8a2f50 [Core] Add MRV2 virtual-batch PCP for MLA (#46570)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
2026-07-19 02:53:15 +00:00
wendadawenandTianyu Guo 69d4f5ef63 [Bugfix][Multimodal] Fix Qwen3-Omni use_audio_in_video with mixed image/video inputs (#46213)
Signed-off-by: wendadawen <wendadawen@qq.com>
Signed-off-by: Tianyu Guo <guoty9@mail2.sysu.edu.cn>
Co-authored-by: Tianyu Guo <guoty9@mail2.sysu.edu.cn>
2026-07-17 08:31:16 +00:00