iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
14,386 Commits 3 Branches 0 Tags
fa6a6be51978bd4b49ba0da17039e60f96dc5b13
Commit Graph
10211 Commits
Author SHA1 Message Date
Zhuohan LiandNick Hill d0cd728907 [Core] Support reseting all running requests' KV while calling reset_prefix_cache (#28827) 
Signed-off-by: Zhuohan Li <zhuohan123@gmail.com>
Signed-off-by: Nick Hill <nhill@redhat.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
 2025-12-02 02:25:05 +00:00
Andrew XiaandAndrew Xia fa8804ad9c [responsesAPI][4] fix responseOutputItem Kimi K2 thinking bug (#29555) 
Signed-off-by: Andrew Xia <axia@fb.com>
Co-authored-by: Andrew Xia <axia@fb.com>
 2025-12-02 02:11:35 +00:00
Divakar Vermaandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> 4b40924998 [ROCm] Fallback pytorch GELU with tanh approximation to GELU() (#29244) 
Signed-off-by: Divakar Verma <divakar.verma@amd.com>
Signed-off-by: Divakar Verma <137818590+divakar-amd@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
 2025-12-02 02:02:22 +00:00
Nick Hill 44822d7ff2 [BugFix] Preserve spec decoding uniform decode when scheduling (#29759) 
Signed-off-by: Nick Hill <nhill@redhat.com>
 2025-12-01 17:15:52 -08:00
Kevin H. Luu 1336a1ea24 Revert #29787 and #29690 (#29815) 2025-12-01 13:42:03 -08:00
Nengjun Ma eaf81485ed [Ascend]: Fixed the issue where OOT Platform vllm-ascend could not enable SP in Eager mode (#28935) 
Signed-off-by: leo-pony <nengjunma@outlook.com>
 2025-12-01 15:02:18 -05:00
shivampr cabc77cc86 [Core][Observability] Add KV cache residency metrics (#27793) 
Introduces three new Prometheus histograms for fine-grained observability of KV cache residency behavior:

vllm:kv_block_lifetime_seconds — total lifetime from allocation to free
vllm:kv_block_idle_before_evict_seconds — idle duration before eviction
vllm:kv_block_reuse_gap_seconds — time between consecutive reuses of the same block

These metrics help operators analyze KV cache efficiency, reuse patterns, and eviction timing beyond simple utilization rates.

Implementation uses monotonic timestamps for accuracy, 1% sampling for minimal overhead (~48 bytes/block), and is fully thread-safe with zero runtime cost when disabled.

Two new runtime flags are introduced:

--kv-cache-metrics – enable KV cache residency metrics
--kv-cache-metrics-sample – control sampling ratio (default: 0.01)

Signed-off-by: Shivam <shivamprasad91@gmail.com>
 2025-12-01 18:27:53 +00:00
knlnguyen1802Chenguang ZhengCyrus LeungCyrus Leung
 		fc6acc88ca [Bugfix] Missing cached item in the MultiModalReceiverCache (#28525) 
Signed-off-by: knlnguyen1802 <knlnguyen1802@gmail.com>
Co-authored-by: Chenguang Zheng <645327136@qq.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
 2025-12-01 10:18:07 -08:00
sangbumlikeagod 092bb73b8a [Frontend] add 'verbose_json' and 'timestamp' feature on Whisper Transcription/Translation (#24209) 
Signed-off-by: sangbumlikeagod <oironese@naver.com>
Signed-off-by: sangbumlikeagod <98077576+sangbumlikeagod@users.noreply.github.com>
 2025-12-01 18:19:17 +01:00
FredericOdermatt 5d43f7372e [Doc] Update description disable_any_whitespace (#29784) 
Signed-off-by: Frederic Odermatt <frederic.odermatt@44ai.ch>
 2025-12-01 16:48:33 +00:00
Shengqi Chen 36db0a35e4 [CI] Renovation of nightly wheel build & generation (#29690) 
Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
 2025-12-01 21:25:39 +08:00
Isotr0pyandbaonudesifeizhai b95db244ee [v1] Add real sliding window calculation to FlexAttention direct BlockMask building (#26015) 
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
Co-authored-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
 2025-12-01 13:12:51 +00:00
Fanli Lin f37e8938d2 [XPU] Fix AWQ skipped layer detection in IPEX quantization (#29774) 
Signed-off-by: Fanli Lin <fanli.lin@intel.com>
 2025-12-01 12:00:52 +00:00
Cyrus Leung f0a28bf661 [Misc] Unify tokenizer registration (#29767) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-12-01 11:34:58 +00:00
Mickaël SeznecandRoger Wang 86e178f7c4 [crashfix] Eagle + multimodal can crash on mm cache miss (#29750) 
Signed-off-by: Mickael Seznec <mickael@mistral.ai>
Co-authored-by: Roger Wang <hey@rogerw.io>
 2025-12-01 17:29:33 +08:00
daniel-salibandChauncey 014ece97c7 [Frontend] Add tool filtering support to ToolServer (#29224) 
Signed-off-by: Daniel Salib <danielsalib@meta.com>
Co-authored-by: Chauncey <chaunceyjiang@gmail.com>
 2025-12-01 08:03:57 +00:00
wang.yuqi 62de4f4257 [Frontend] Resettle pooling entrypoints (#29634) 
Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>
 2025-12-01 15:30:43 +08:00
Yifei Zhang 1ab8fc8197 Make PyTorch profiler gzip and CUDA time dump configurable (#29568) 
Signed-off-by: Yifei Zhang <yifei.zhang1992@outlook.com>
 2025-12-01 04:30:46 +00:00
Shu WangrootMichael Goin
 		f72a817bdf [MoE] CuteDSL MoE with Nvfp4 DeepEP dispatch (#27141) 
Signed-off-by: Shu Wang <shuw@nvidia.com>
Signed-off-by: Shu Wang. <shuw@nvidia.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: root <root@umbriel-b200-017.ipp4a1.colossus.nvidia.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
 2025-11-30 16:05:32 -08:00
Woosuk Kwon ec38a7368d [Model Runner V2] Use packed mask for prompt bin counts (#29756) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-30 14:15:42 -08:00
Xingyu LiuandHarry Mellor 21c2627934 [Misc]Remove redundant hidden_size property in ModelConfig (#29749) 
Signed-off-by: Xingyu Liu <charlotteliu12x@gmail.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
 2025-11-30 17:14:23 +00:00
Omer Ullman Argov 39d28108f4 [Feat] Support non-gated activations in NVFP4 modelopt path (#29004) 2025-11-30 11:02:40 -05:00
Harry Mellor cd719de5cb Fix RoPE failures in Transformers nightly (#29700) 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
 2025-11-30 14:29:32 +00:00
Pleaplusone 8c363ed666 [ROCm][Attention] Sliding window support for AiterFlashAttentionBackend (#29234) 
Signed-off-by: ganyi <ygan@amd.com>
 2025-11-30 11:31:50 +00:00
Cyrus Leung 64bc09ba27 [Core] Enable inputs_embeds_size separate from hidden_size (#29741) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-11-30 17:31:12 +08:00
Isotr0py 47539cfd3e [Bugfix] Fix mismatched nvfp4 gemm output shape (#29742) 
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
 2025-11-30 09:15:01 +00:00
Cyrus Leung 2afcec4dec [Misc] Update TokenizerLike interface and move get_cached_tokenizer (#29730) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-11-30 14:59:47 +08:00
朝 9381b5cde0 [Doc]: Fix typo in fused_moe layer (#29731) 
Signed-off-by: BowTen <bowten@qq.com>
 2025-11-29 22:29:13 -08:00
VensenandTJian 66b5840287 [Bugfix][sleepmode][fp8 kv cache]: Fix FP8 KV cache + sleep(level=2) gibberish output (#28783) 
Signed-off-by: vensen <vensenmu@gmail.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
 2025-11-30 14:24:25 +08:00
Huamin Li 82c795d6f2 Fix AttributeError about _use_fi_prefill (#29734) 
Signed-off-by: Huamin Li <3ericli@gmail.com>
 2025-11-30 06:04:55 +00:00
Isotr0py e1464c3a08 [Quantization] Enable compressed-tensors AWQ for Turing GPU (#29732) 
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
 2025-11-30 06:04:28 +00:00
Xin YangandJee Jee Li a491b0911b [LoRA] Support FusedMoE LoRA Triton kernel for mxfp4 (#29708) 
Signed-off-by: Xin Yang <xyangx@amazon.com>
Signed-off-by: Xin Yang <105740670+xyang16@users.noreply.github.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
 2025-11-30 10:37:25 +08:00
Jinzhen LinMichael GoinMichael Goin
 		1656ad3704 [Kernel][Quantization] add w4a8 support for marlin kernel (#24722) 
Signed-off-by: Jinzhen Lin <jinzhen.ljz@antgroup.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Michael Goin <mgoin@redhat.com>
 2025-11-29 07:19:33 -08:00
Cyrus Leung fa59fe417f [Chore] Move detokenizer_utils to vllm/tokenizers (#29727) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-11-29 06:25:17 -08:00
Cyrus Leung fe3398fab2 [Chore] Enable passing tokenizer=None into MM processor (#29724) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-11-29 06:25:10 -08:00
Cyrus Leung 34a984274e [Misc] Refactor tokenizer interface (#29693) 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
 2025-11-29 04:02:21 -08:00
Woosuk Kwon f223ed4181 [Model Runner V2] Fuse penalties and temperature into single kernel (#29720) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-29 02:29:16 -08:00
Didier Durand 04a797cd0e [Doc]: fixing typos in various files. (#29717) 
Signed-off-by: Didier Durand <durand.didier@gmail.com>
 2025-11-29 01:15:39 -08:00
Woosuk Kwon 6afc0ffaf6 [Model Runner V2] Add sample/ directory and reorganize files (#29719) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-29 00:41:01 -08:00
Jee Jee LiandCyrus Leung 39e63dec7c [LoRA] Cleanup LoRA unused code (#29611) 
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
 2025-11-28 22:52:58 -08:00
Woosuk Kwon 4a80ad0a25 [Model Runner V2] Don't use UVA buffer for prefill_len (#29713) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-28 20:27:16 -08:00
Lucas Wilkinson e23f665d83 [BugFix] Fix DBO failing with TypeError: 'NoneType' object is not iterable (#29698) 
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
 2025-11-28 20:19:01 -08:00
Woosuk Kwon ca1b1e7296 [Model Runner V2] Refactor prefill token preparation (#29712) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-28 19:49:17 -08:00
Tsukasa OI 762a4a6ca9 [Frontend] Perform offline path replacement to tokenizer (#29706) 
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
 2025-11-28 18:32:08 -08:00
Woosuk Kwon 1dcafb3dea [Model Runner V2] Support penalties using bin counts (#29703) 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
 2025-11-28 17:53:17 -08:00
Mert UnsalandRoger Wang c625d7b1c6 [Bugfix] Fix O(n²) multimodal string prompt processing (#29667) 
Signed-off-by: mertunsall <mertunsal1905@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
 2025-11-28 16:10:39 -08:00
Zhengxu Chen 6173682b6e [compile] Include enable_sleep_mode into caching factors. (#29696) 
Signed-off-by: zhxchen17 <zhxchen17@fb.com>
 2025-11-29 07:58:38 +08:00
Augusto Yao 9726e64530 bugfix: correct attn output with base 2 or e (#28840) 
Signed-off-by: augusto.yjh <augusto.yjh@antgroup.com>
 2025-11-29 07:52:12 +08:00
Huamin Li 3fd1fb0b60 Revert "[LoRA] Support FusedMoE LoRA Triton kernel for mxfp4 (#28971)" (#29697) 
Signed-off-by: Huamin Li <3ericli@gmail.com>
 2025-11-28 15:26:52 -08:00
Jiangyun ZhuandRoger Wang a51f4186f2 [Bugfix] fix dots.llm1.inst (#29687) 
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
 2025-11-28 15:25:26 -08:00