Skip to content

Merge with main #99

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 185 commits into
base: sage/dbo-full-cudagraphs
Choose a base branch
from
Open

Merge with main #99

wants to merge 185 commits into from

Conversation

yewentao256
Copy link

@yewentao256 yewentao256 commented Aug 20, 2025 •
edited by github-actions bot
Loading

Purpose

Test Plan

Test Result

(Optional) Documentation Update

Essential Elements of an Effective PR Description Checklist
  • The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
  • The test plan, such as providing test command.
  • The test results, such as pasting the results comparison before and after, or e2e results
  • (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

huachenheli and others added 30 commits August 15, 2025 18:23
@huachenheli
[Benchmarks] Include image data when ShareGPT4V dataset is used. (vll…
993d3d1 
…m-project#22955)

Signed-off-by: Chenheli Hua <[email protected]>
@shen-shanshan
[Structured Output] Make the output of structured output example more…
48b01fd 
… complete (vllm-project#22481)

Signed-off-by: shen-shanshan <[email protected]>
@bnellnm @mgoin
[Kernels] Clean up FusedMoeMethodBase and modular kernel setup. Remov…
8ad7285 
…e extra arguments from modular kernel methods. (vllm-project#22035)

Signed-off-by: Bill Nell <[email protected]>
Co-authored-by: Michael Goin <[email protected]>
@cyang49
[Model] Granite-4 support loading quantized checkpoint (vllm-project#…
6cd69f5 
…22925)

Signed-off-by: Chih-Chieh-Yang <[email protected]>
@yewentao256
[Log] Debug Once for Randomizing dummy data for DP Rank (vllm-project…
df5afa8 
…#22860)

Signed-off-by: yewentao256 <[email protected]>
@linzebing
[Core] direct indexing on self.block_table_np in compute_slot_mapping (…
6e67077 
…vllm-project#22940)

Signed-off-by: linzebing <[email protected]>
@nvjullin
[Bugfix] Added more env vars to hash (vllm-project#22449)
79899b6 
Signed-off-by: Julien Lin <[email protected]>
@mgoin
Use regex in convert-results-json-to-markdown.py (vllm-project#22989)
a344a1a 
Signed-off-by: Michael Goin <[email protected]>
@mgoin
[CI] Speed up Whisper tests by reusing server (vllm-project#22859)
8a87cd2 
Signed-off-by: mgoin <[email protected]>
@shixianc
[Fix] enable swap_ab for pplx problem size computation (vllm-project#…
7f89ed2 
…22991)

Signed-off-by: Shixian Cui <[email protected]>
Co-authored-by: Shixian Cui <[email protected]>
@eicherseiji
Add PrefixRepetitionRandomDataset to vllm bench serve datasets (vll…
00d6cba 
…m-project#20638)

Signed-off-by: Seiji Eicher <[email protected]>
@yyihuang
minor: zero workspace buffer init for flashinfer trtllm-gen attn (vll…
1723ef1 
…m-project#22603)
@LucasWilkinson
[Attention] FA3 Attention Sinks Perf Boost (vllm-project#22478)
177e55e 
Signed-off-by: Lucas Wilkinson <[email protected]>
@tdoublep
[BugFix] Fix regression caused by mamba state dtype PR (vllm-project#…
f5d412b 
…22998)

Signed-off-by: Thomas Parnell <[email protected]>
@seemethere
ci: Add CUDA + arm64 release builds (vllm-project#21201)
76144ad 
Signed-off-by: Eli Uriegas <[email protected]>
@rishitdholakia13
[Structured Outputs] [Bug] Fix misalignment in apply_grammar_bitmask …
1fc375d 
…causing unintended masking and NaN logits (vllm-project#22963)

Signed-off-by: rishitdholakia13 <[email protected]>
@njhill @yinghai
[BugFix] Handle case where async utility call is cancelled (vllm-proj…
b9dc9d2 
…ect#22996)

Signed-off-by: Nick Hill <[email protected]>
Co-authored-by: Yinghai Lu <[email protected]>
@orozery
[v1] Move block_hashes from KVCacheManager to Request.block_hashes (v…
c280066 
…llm-project#19728)

Signed-off-by: Or Ozeri <[email protected]>
@sarckk
Support multiple attention groups for KV sharing (vllm-project#22672)
3e2f798 
Signed-off-by: Yong Hoon Shin <[email protected]>
@oraluben
[BugFix] Make run_once thread-safe (vllm-project#22978)
236b864 
Signed-off-by: <[email protected]>
Signed-off-by: Yichen Yan <[email protected]>
@njhill
[Misc] Support passing multiple request ids at once to `AsyncLLM.abor…
ad0297d 
…t()` (vllm-project#22944)

Signed-off-by: Nick Hill <[email protected]>
@NickLucche
[Kernel] Simplify get_kv_cache_layout and cache `use_trtllm_attenti…
070da66 
…on` env-dependent bit (vllm-project#22735)

Signed-off-by: NickLucche <[email protected]>
@benchislett
[Bugfix] Fix DeepSeek MTP (vllm-project#22934)
fbd8872 
Signed-off-by: Benjamin Chislett <[email protected]>
@njhill
[Frontend] Avoid list copies in serving_chat.py (vllm-project#22947)
f6b5040 
Signed-off-by: Nick Hill <[email protected]>
@calvin0327 @njhill
[V1] support min_tokens for detokener (vllm-project#22014)
e4e37de 
Signed-off-by: calvin chen <[email protected]>
Co-authored-by: Nick Hill <[email protected]>
@gracehonv
[misc] nsys profile output kernel classifier and visualizer (vllm-pro…
1f83e7d 
…ject#22971)

Signed-off-by: Grace Ho <[email protected]>
@jikunshang
[XPU]avoid circular import during XPU init (vllm-project#23017)
7caec10 
Signed-off-by: Kunshang Ji <[email protected]>
@LucasWilkinson
[Build] Env var to disable sccache (vllm-project#22968)
5157827 
Signed-off-by: Lucas Wilkinson <[email protected]>
@qthequartermasterman
[BugFix] Add support for loading prompt embeds tensors serialized on …
78863f8 
…unavailable devices and sparse tensors (vllm-project#22962)

Signed-off-by: Andrew Sansom <[email protected]>
@jeejeelee
[Misc] Add --save-dir option to benchmark_moe (vllm-project#23020)
6d3da47 
Signed-off-by: Jee Jee Li <[email protected]>
MatthewBonanni and others added 30 commits August 20, 2025 21:43
@MatthewBonanni
Remove chunked_prefill_enabled flag in V1 MLA (vllm-project#23183)
a4fbb32 
Signed-off-by: Matthew Bonanni <[email protected]>
@MatthewBonanni
Feature/mla tests (vllm-project#23195)
10cc12b 
Signed-off-by: Matthew Bonanni <[email protected]>
Signed-off-by: Matthew Bonanni <[email protected]>
@shixianc
[Fix] remove is_marlin param in benchmark_moe (vllm-project#23286)
c86af22
@22quinn @simon-mo
[EP] Add logging for experts map (vllm-project#22685)
4b79502 
Signed-off-by: 22quinn <[email protected]>
Co-authored-by: Simon Mo <[email protected]>
@russellb
Remove duplicate entry in vllm.attention.__all__ (vllm-project#23296)
f5aa307 
Signed-off-by: Russell Bryant <[email protected]>
@mgoin
[CI Bugfix] Fix CI by fully removing --enable-prompt-adapter (vllm-pr…
bbea1ce 
…oject#23284)

Signed-off-by: mgoin <[email protected]>
@WoosukKwon
[Optimization] Make new_block_ids None if empty (vllm-project#23262)
b029de9 
Signed-off-by: Woosuk Kwon <[email protected]>
@bigPYJ1151
[CPU] Refactor CPU W8A8 scaled_mm (vllm-project#23071)
7be5d11 
Signed-off-by: jiang1.li <[email protected]>
@DarkLight1337
[CI/Build] Split out mm processor tests (vllm-project#23260)
2461d9e 
Signed-off-by: DarkLight1337 <[email protected]>
@Josephasafg @asafgardin
[V1][Mamba1] - Full CUDA and Piecewise CUDA Graphs Support (vllm-proj…
3663870 
…ect#23035)

Signed-off-by: asafg <[email protected]>
Signed-off-by: asafg <[email protected]>
Co-authored-by: asafg <[email protected]>
@yewentao256
[Compile] Fix Compile Warning SM100 Cutlass MLA (vllm-project#23287)
f94bf9b 
Signed-off-by: yewentao256 <[email protected]>
@yannqi @gemini-code-assist @DarkLight1337
[Model][VLM] Support R-4B Model (vllm-project#23246)
655a09f 
Signed-off-by: yannqi <[email protected]>
Signed-off-by: 杨奇(yann qi) <[email protected]>
Signed-off-by: Cyrus Leung <[email protected]>
Co-authored-by: yannqiyang <[email protected]>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Cyrus Leung <[email protected]>
@QiliangCui
[CI] Delete images older than 24h. (vllm-project#23291)
8993073 
Signed-off-by: Qiliang Cui <[email protected]>
@mgoin
[CI] Block the cu126 wheel build while broken (vllm-project#23285)
f64ee61 
Signed-off-by: mgoin <[email protected]>
@22quinn @njhill @WoosukKwon
[Sampler] Support returning final logprobs (vllm-project#22387)
f571ff8 
Signed-off-by: 22quinn <[email protected]>
Co-authored-by: Nick Hill <[email protected]>
Co-authored-by: Woosuk Kwon <[email protected]>
@DarkLight1337
[Bugfix] Fix extra whitespace in strings caused by newline (vllm-proj…
0c31e28 
…ect#23272)

Signed-off-by: DarkLight1337 <[email protected]>
@jaredoconnell @DarkLight1337
[BugFix] Fix Python 3.9 Support (vllm-project#23306)
3128240 
Signed-off-by: Jared O'Connell <[email protected]>
Signed-off-by: Cyrus Leung <[email protected]>
Co-authored-by: Cyrus Leung <[email protected]>
@paulpak58
[Model] Add LFM2 architecture (vllm-project#22845)
2e2000f 
Signed-off-by: Paul Pak <[email protected]>
@DarkLight1337
[Refactor] Simplify code for MM budget (vllm-project#23310)
0c6e40b 
Signed-off-by: DarkLight1337 <[email protected]>
@DarkLight1337 @youkaichao
[Doc] Fix batch-level DP example (vllm-project#23325)
5cc54f7 
Signed-off-by: DarkLight1337 <[email protected]>
Signed-off-by: Cyrus Leung <[email protected]>
Co-authored-by: youkaichao <[email protected]>
@noooop
[Performance] V1 Pooling Models E2E Performance Optimization (vllm-pr…
d70a166 
…oject#23162)

Signed-off-by: wang.yuqi <[email protected]>
@robertgshaw2-redhat
[V1] Remove unnecessary check for main thread (vllm-project#23298)
c8e33c7 
Signed-off-by: Robert Shaw <[email protected]>
Co-authored-by: Robert Shaw <[email protected]>
@zhuangqh
[Bugfix] set system_message in phi4mini chat template (vllm-project#2…
f8daddc 
…3309)

Signed-off-by: zhuangqh <[email protected]>
@ywang96 @DarkLight1337
[Multimodal] Always enable hashing mm data (vllm-project#23308)
79f05e4 
Signed-off-by: Roger Wang <[email protected]>
Signed-off-by: DarkLight1337 <[email protected]>
Co-authored-by: DarkLight1337 <[email protected]>
@youkaichao
[ci/build] Fix abi tag for aarch64 (vllm-project#23329)
e0b056e 
Signed-off-by: youkaichao <[email protected]>
@bbeckca
Migrate MolmoImageInputs to TensorSchema (vllm-project#22022)
a482e4e 
Signed-off-by: Benji Beck <[email protected]>
@yiliu30 @yewentao256
Fix nvfp4 swizzling (vllm-project#23140)
0278f1a 
Signed-off-by: yiliu30 <[email protected]>
Co-authored-by: Wentao Ye <[email protected]>
@IwakuraRein @mgoin
add tg-mxfp4-moe-test (vllm-project#22540)
f8ce022 
Signed-off-by: siyuanf <[email protected]>
Signed-off-by: Siyuan Fu <[email protected]>
Co-authored-by: Michael Goin <[email protected]>
@yewentao256 @mgoin
[Bug] Fix R1 Accuracy 0 Bug (vllm-project#23294)
48bfb0c 
Signed-off-by: yewentao256 <[email protected]>
Signed-off-by: Wentao Ye <[email protected]>
Co-authored-by: Michael Goin <[email protected]>
@yewentao256
Merge branch 'main' into wye-dbo-full-cuda-graph
db21144 
Signed-off-by: yewentao256 <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tlrmchlsmth tlrmchlsmth Awaiting requested review from tlrmchlsmth tlrmchlsmth is a code owner

@robertgshaw2-redhat robertgshaw2-redhat Awaiting requested review from robertgshaw2-redhat robertgshaw2-redhat is a code owner

@njhill njhill Awaiting requested review from njhill njhill is a code owner

@alexm-redhat alexm-redhat Awaiting requested review from alexm-redhat alexm-redhat is a code owner

@youkaichao youkaichao Awaiting requested review from youkaichao youkaichao is a code owner

@mgoin mgoin Awaiting requested review from mgoin mgoin is a code owner

@ProExpertProg ProExpertProg Awaiting requested review from ProExpertProg ProExpertProg is a code owner

@LucasWilkinson LucasWilkinson Awaiting requested review from LucasWilkinson LucasWilkinson is a code owner

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.