Skip to content

Navigation Menu

Appearance settings

- AI CODE CREATION
  GitHub CopilotWrite better code with AI
  GitHub Copilot appDirect agents from issue to merge
  MCP Registry^NewIntegrate external tools
- DEVELOPER WORKFLOWS
  ActionsAutomate any workflow
  CodespacesInstant dev environments
  IssuesPlan and track work
  Code ReviewManage code changes
- APPLICATION SECURITY
  GitHub Advanced SecurityFind and fix vulnerabilities
  Code securitySecure your code as you build
  Secret protectionStop leaks before they start
- EXPLORE
  Why GitHub
  Documentation
  Blog
  Changelog
  Marketplace
View all features
- BY COMPANY SIZE
  Enterprises
  Small and medium teams
  Startups
  Nonprofits
- BY USE CASE
  App Modernization
  DevSecOps
  DevOps
  CI/CD
  View all use cases
- BY INDUSTRY
  Healthcare
  Financial services
  Manufacturing
  Government
  View all industries
View all solutions
- EXPLORE BY TOPIC
  AI
  Software Development
  DevOps
  Security
  View all topics
- EXPLORE BY TYPE
  Customer stories
  Events & webinars
  Ebooks & reports
  Business insights
  GitHub Skills
- SUPPORT & SERVICES
  Documentation
  Customer support
  Community forum
  Trust center
  Partners
View all resources
- COMMUNITY
  GitHub SponsorsFund open source developers
- PROGRAMS
  Security Lab
  Maintainer Community
  Accelerator
  GitHub Stars
  Archive Program
- REPOSITORIES
  Topics
  Trending
  Collections
- ENTERPRISE SOLUTIONS
  Enterprise platformAI-powered developer platform
- AVAILABLE ADD-ONS
  GitHub Advanced SecurityEnterprise-grade security features
  Copilot for BusinessEnterprise-grade AI features
  Premium SupportEnterprise-grade 24/7 support
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

Appearance settings

You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

Uh oh!

There was an error while loading. Please reload this page.

dphnAI / sonar Public

Notifications You must be signed in to change notification settings
Fork 205
Star 1.8k

Code
Issues 77
Pull requests 49
Discussions
Actions
Projects
Wiki
Security and quality
Insights

Additional navigation options

Code
Issues
Pull requests
Discussions
Actions
Projects
Wiki
Security and quality
Insights

[sync] vLLM 2026-07-15 — e26264f3e..647213129#1710

Merged

AlpinDale merged 77 commits into

maindphnAI/sonar:mainfrom

sync/vllm-2026-07-15-647213129dphnAI/sonar:sync/vllm-2026-07-15-647213129Copy head branch name to clipboard

Jul 16, 2026

Conversation Commits77 (77)Checks Files changed

Merged

[sync] vLLM 2026-07-15 — e26264f3e..647213129#1710
AlpinDale merged 77 commits into
mainfrom
sync/vllm-2026-07-15-647213129

Commits

Commits on Jul 15, 2026

[sync] [Kernel] Implement CUDA kernel for ReLUSquaredActivation (relu^2) (#39058)

AlpinDale
and
tanish-malekar
committed
[sync] [ROCm][MiniMax-M3] Add AITER sparse paged attention (#47287)

AlpinDale
and
tanpinsiang
committed
[sync] [Model] Optimize Qwen3.5 on H20 (#48350)

AlpinDale
and
zzt93
committed
[sync] [Bugfix][ROCm] Keep TP all_gather on base-class collective (#48446)

AlpinDale
and
Fangzhou-Ai
committed
[sync] Re-disable CUDA graph memory profiling on ROCm (#48440)

AlpinDale
and
Rohan138
committed
[sync] [CPU][Spec Decode] Support DFlash speculative decoding for GDN models on CPU (#46090)

AlpinDale
and
guybd
committed
[sync] [Bugfix][UT]Fix EagleMiniCPMForCausalLM meet TypeError (#48452)

AlpinDale
and
Yejing-Lai
committed
[sync] [CPU] Create Proper Numa topology for s390x (#40714)

AlpinDale
and
R3hankhan123
committed
[sync] [BugFix] Restore full tokens for Qwen MTP When MoE SP (#48429)

AlpinDale
and
gcanlin
committed
[sync] [ROCm][CI] Cache Rust builds by source inputs (#46527)

AlpinDale
and
AndreasKaratzas
committed
[sync] [Distributed][Perf] Enable FlashInfer MNNVL allreduce RMS quant fusion (#48064)

AlpinDale
and
mmangkad
committed
[sync] [Attention] Make sliding-window support an explicit backend capability (#48011)

AlpinDale
and
NickLucche
committed
[sync] [Bugfix][KV Cache] Don't route uniform-page-size MLA+SWA models into DeepseekV4 packing (#48256)

AlpinDale
and
NickLucche
committed
[sync] [Core] Support fp32 lm_head for generation models via head_dtype (RFC #48305 §3.6) (#48390)

AlpinDale
and
KKothuri
committed
[sync] [CI] Add SPDX license header to Rust/Protobuf sources (#48472)

AlpinDale
and
BugenZhao
committed
[sync] [Mypy Fix] Split mypy work (#48490)

AlpinDale
and
yewentao256
committed
[sync] lower memory required for capturing cudagraphs for large cudagraph sizes (#48483)

AlpinDale
and
omera-nv
committed
[sync] remove force channels_last in Idefics3MultiModalProcessor (#48467)

AlpinDale
and
yma11
committed
[sync] [BugFix][ModelRunner V2] Fix stale attn metadata in speculator prefill cudagraph capture (#48261)

AlpinDale
and
njhill
committed
[sync] [Misc] Improve Matryoshka pooling dimensions validation (#48057)

AlpinDale
and
taneem-ibrahim
committed
[sync] [EC Connector] CPU Offloading EC Connector (#47423)

AlpinDale
and
omerpaz95
committed
[sync] [ROCm][CI] Transformers: pass only one of input_ids/inputs_embeds (#48258)

AlpinDale
and
stefankoncarevic
committed
[sync] [Doc] Add DeepseekV32ForCausalLM to supported_models.md (#48293)

AlpinDale
and
Gavin-Morris-04
committed
[sync] [Bugfix][Frontend] Flush engine reasoning parser at engine-reasoning → tool streaming boundary (#47606)

AlpinDale
and
akii96
committed
[sync] [ROCm][CI] Unblock `AMD: Language Models Test (Extended Pooling)` (#48513)

AlpinDale
and
micah-wil
committed
[sync] Add DCP + Eagle support for Tokenspeed MLA backends (#48180)

AlpinDale
and
pavanimajety
committed
[sync] Added sliding window attention support for qwen-eagle3 architecture (#47568)

AlpinDale
and
shanjiaz
committed
[sync] [Core] Preserve Marconi caching with selective hybrid cache retention (#47782)

AlpinDale
and
njhill
committed
[sync] [BugFix] Initialize model_config for Qwen3-VL MoE (#44863)

AlpinDale
and
wenpengw-nv
committed
[sync] [Feat] Add Support for BertForMaskedLM to vLLM (#48463)

AlpinDale
and
atalhens
committed
[sync] [Frontend] Expose logprob_token_ids on Python OpenAI endpoints (#43463)

AlpinDale
and
langzhao-netizen
committed
[sync] [BugFix] Correct OTEL span start time for Dynamo compilation (#40698)

AlpinDale
and
emricksini-h
committed
[sync] add pad-aware swiglu limit kernel (#48287)

AlpinDale
and
gnovack
committed
[sync] [Quantization][INC][ARK] Support INT2 XPU WOQ Linear (#47521)

AlpinDale
and
Zhenzhong1
committed
[sync] [1/N] Add dense MHA path for sparse MLA short sequences (#47327)

AlpinDale
and
MatthewBonanni
committed
[sync] up FI fp8 moe topk to 32 (#44462)

AlpinDale
and
DanBlanaru
committed
[sync] [ROCm][MiniMax-M2] Dispatch fused QK-norm + AllReduce via AITER (#44849)

AlpinDale
and
akii96
committed
[sync] [ROCm][Kernel] Add HybridW4A16LinearKernel: Triton prefill + HIP skinny decode (#40977)

AlpinDale
and
mgehre-amd
committed
[sync] [Perf][ROCm] Fix GDN KKT warmup regression on RDNA by avoiding fp32 tl.dot (#45000)

AlpinDale
and
nemanjaudovic
committed
[sync] [ROCm][MiniMax-M3][Spec Decode] Support speculative decode with AITER sparse PA (#47984)

AlpinDale
and
tanpinsiang
committed
[sync] [Bugfix] Skip minimax_m3 tool parser tests when Rust extension is absent (#48523)

AlpinDale
and
mwoodson
committed
[sync] [Misc] Clean up "swap_space" (#48549)

AlpinDale
and
wangxiyuan
committed
[sync] [ROCm] Retune MI355 selective_state_update float16 config on the unified effective_batch grid (#48372)

AlpinDale
and
vanshbhatia-amd
committed
[sync] [Perf][Feat] Add generic cuteDSL LL BF16 router (GEMM) (#42562)

AlpinDale
and
LopezCastroRoberto
committed
[sync] fix(security): guard lm-format-enforcer regex compile with timeout (#47595)

AlpinDale
and
jperezdealgaba
committed
[sync] Add Cosmos3 Edge Reasoner model (#48291)

AlpinDale
and
adsridhar
committed
[sync] [Bugfix] Return 400 instead of 500 when multimodal data is sent to a text-only model (#48473)

AlpinDale
and
hnt2601
committed
[sync] [Test] Enable KV cache events for HMA models in CPU offloading test (#47754)

AlpinDale
and
Etelis
committed
[sync] [Misc] Rename VLLM_TRITON_ATTN_USE_TD to VLLM_TRITON_USE_TD (#45781)

AlpinDale
and
afierka-intel
committed
[sync] [Bugfix][Security] Fix concurrent sparse invariant race bypassing CVE remediation (#48583)

AlpinDale
and
jperezdealgaba
committed
[sync] [Model] Enable LoRA support for tower and connector in LlavaNextVideo (#48594)

AlpinDale
and
gangula-karthik
committed
[sync] [NIXL] Avoid reading expired blocks in bidirectional turn-2 read (#47021)

AlpinDale
and
tomerg-nvidia
committed
[sync] [Bugfix] Include inline per-token-head scales in offloaded page transfer width (#48411)

AlpinDale
and
Etelis
committed
[sync] [Bugfix] Gemma4 parser: classify channel-less output consistently in streaming and non-streaming (#48262)

AlpinDale
and
adhi29
committed
[sync] [CI Bug] Fully solve accuracy issue for DSv3.2 + MTP + Sequence Parallel (#48036)

AlpinDale
and
yewentao256
committed
[sync] [Bugfix] Make MLA+SWA check the layer's backend, not the model config (#48520)

AlpinDale
and
mgoin
committed
[sync] [Doc] Sync four function docstrings with their signatures (#45437)

AlpinDale
and
DaoyuanLi2816
committed
[sync] [KV Offload] Split cpu_cache_usage_perc into write/read usage gauges (#47666)

AlpinDale
and
Srinivasoo7
committed
[sync] [ROCm] Retune MI355 selective_state_update float32 config on the unified effective_batch grid (#48373)

AlpinDale
and
vanshbhatia-amd
committed
[sync] [Reasoning] Optimize TPOT for thinking budget when used with speculative decoding (#46662)

AlpinDale
and
rishitdholakia13
committed
[sync] [CI] Build macOS arm64 CPU wheel natively on the macmini queue (#48289)

AlpinDale
and
mgoin
committed
[sync] [ROCm][CI] fix flashinfer import check (#48647)

AlpinDale
and
divakar-amd
committed
[sync] [CI][Bugfix] Fix FlashAttention reported MLA dimension support (#48631)

AlpinDale
and
MatthewBonanni
committed
[sync] Log fully resolved pooling config at startup (#48030)

AlpinDale
and
taneem-ibrahim
committed
[sync] [Bugfix][CI] Fix test_head_dtype quant_method test on ROCm (#48654)

AlpinDale
and
micah-wil
committed
[sync] fix: size FlashInfer prefill workspace to batch head footprint (#48428)

AlpinDale
and
joerowell
committed
[sync] [Bugfix][R3] Exclude draft routers from expert capture (#48622)

AlpinDale
and
aoshen02
committed
[sync] [Bugfix] Preserve unloaded non-persistent buffers during layerwise reload (#44371)

AlpinDale
and
joanvelja
committed
[sync] [Perf] Remove redundant repeat and copy for dsv4, 1.8% E2E TPOT improvement. (#48137)

AlpinDale
and
yewentao256
committed
[sync] add pad-aware reduce path (#48385)

AlpinDale
and
gnovack
committed
[sync] [ROCm][CI] Remove mxfp4 test skips after `amd-quark` 0.12 release (#47330)

AlpinDale
and
micah-wil
committed
[sync] [Core][LoRA] Support fp32 lm_head (head_dtype) on the LoRA path (#48525)

AlpinDale
and
KKothuri
committed
[sync] [Quant] Enable humming w[2-7]a[4,8] inference with compressed-tensors (#46390)

AlpinDale
and
HDCharles
committed
[sync] [LoRA][1/N] Integrate flashinfer MoE LoRA for BF16 model (#48632)

AlpinDale
and
jeejeelee
committed
[sync] [Security] Replace diskcache to eliminate pickle deserialization (#44549)

AlpinDale
and
russellb
committed
[sync] Build with ABI stable FlashMLA (#48174)

AlpinDale
and
janeyx99
committed
[sync] [Bugfix] Set kv_quant_mode on the generic MLA KV-cache spec (#48379)

AlpinDale
and
drakosha
committed

Footer

© 2026 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Community
Docs
Contact

You can’t perform that action at this time.