Repository navigation

[Core] Align vLLM-Omni with vLLM v0.32.0 - #9066

Merged

hsliuustc0106 merged 63 commits into

vllm-project:mainfrom

tzhouam:dev/vllm-align-v0320-main-20261019

Oct 19, 2026

Merged

[Core] Align vLLM-Omni with vLLM v0.32.0#9066
hsliuustc0106 merged 63 commits into
vllm-project:mainfrom
tzhouam:dev/vllm-align-v0320-main-20261019

Commits on Oct 1, 2026
- ci: pin the official vLLM main commit image
  tzhouam
  committed
  383cc5e
Commits on Oct 2, 2026
- rebase: align vllm-omni with vLLM ac9126e58aa7
  tzhouam
  committed
  2721f16
- rebase: align vllm-omni with vLLM ac9126e58aa7
  tzhouam
  committed
  0c84ede
Commits on Oct 3, 2026
- fix: preserve Omni processor conditioning on vLLM 0.31
  tzhouam
  committed
  84cbf94
- Merge main into dev/vllm-align for vLLM 0.31 release
  tzhouam
  committed
  008fa25
- rebase: target the official vLLM 0.31.0 release
  tzhouam
  committed
  ed876de
- test: use the vLLM 0.31 model-length property in graph fixtures
  tzhouam
  committed
  2974f8b
- fix: match vLLM 0.31 generation profiling API
  tzhouam
  committed
  16f73d0
- fix: align MRV2 profiling and capture with vLLM 0.31
  tzhouam
  committed
  43e05c6
- fix: scope CI package version overrides to Omni
  tzhouam
  committed
  c8d5f99
- test: update the Moss runner fixture for release output APIs
  tzhouam
  committed
  864bc7b
- test: fit Qwen2.5 online expansion on its declared two GPUs
  tzhouam
  committed
  0b6591a
- Adapt Moss sampling and runner fixtures to vLLM 0.31 interfaces
  tzhouam
  committed
  b256e1e
- Honor Omni stage runner selection in shared vLLM config
  tzhouam
  committed
  9361c52
- Fit weekly serving fixtures to assigned GPUs and initialize accuracy groups
  tzhouam
  committed
  649b0f7
- Use blockwise Flux offload within single-L4 memory capacity
  tzhouam
  committed
  0dd5f3e
- Leave Code2Wav workspace headroom in two-L4 serving fixture
  tzhouam
  committed
  757dd3b
- Update realtime invalid-request fixtures and reserve Moss encoder memory
  tzhouam
  committed
  37803a8
- Format realtime reliability fixtures
  tzhouam
  committed
  cb8d9f2
- Provide Omni lifecycle state for auxiliary v2 stages
  tzhouam
  committed
  aa499fe
- Keep release alignment focused and clarify baseline fixes
  tzhouam
  committed
  df3807e
- Synchronize CPU ring workers before group teardown
  tzhouam
  committed
  4ad71d8
Commits on Oct 4, 2026
Commits on Oct 5, 2026
- Drop CosyVoice regression changes superseded by main
  tzhouam
  committed
  df14622
- Align source installation documentation with vLLM 0.31.0
  tzhouam
  committed
  07326d0
Commits on Oct 18, 2026
- ci: pin the CI image to an unreleased vLLM main commit
  
  tzhouam
  and
  claude
  committed
  287b418
- rebase: align vllm-omni with vLLM e0317b28ce3b
  tzhouam
  committed
  e6b8f46
- Merge origin/dev/vllm-align (v0.31 campaign line) into the v0.32 alignment
  tzhouam
  committed
  3c6f95e
- ci: re-pin the CI base image to a published nightly
  tzhouam
  committed
  a91d870
- rebase: align vllm-omni with vLLM v0.32.0rc2
  tzhouam
  committed
  e55940e
- rebase: align vllm-omni with vLLM v0.32.0rc3
  tzhouam
  committed
  f107aa4
- docs: align source installation with vLLM 0.32 release branch
  tzhouam
  committed
  59c1a0f
- [Core] Align streaming and handshake fixtures with vLLM 0.32
  tzhouam
  committed
  114c90a
- [Core] Align with verified vLLM 0.32 release wheel
  tzhouam
  committed
  b0d3b30
- [CI/Build] Support config-based vision RoPE in accuracy reference
  tzhouam
  committed
  fd73689
- [Doc] Reflect published vLLM 0.32 release
  tzhouam
  committed
  cd85678
- [Doc] Record published vLLM 0.32 platform artifacts
  tzhouam
  committed
  01f2882
- [CI/Build] Adopt published vLLM 0.32 release images
  tzhouam
  committed
  3d38f0a
- [Bugfix] Align video option validation with vLLM 0.32
  tzhouam
  committed
  00ac6b3
- [CI/Build] Report CUDA runtime dependency versions
  tzhouam
  committed
  93e641c
- [CI/Build] Align orchestrator video fixture with vLLM 0.32
  tzhouam
  committed
  307faa9
- [CI/Build] Add opt-in L4 alignment diagnostics
  tzhouam
  committed
  becbf7d
- [Bugfix] Preserve multimodal encoder attention backend overrides
  tzhouam
  committed
  f723682
- [CI/Build] Validate diagnostic config ownership and compare audio decode
  tzhouam
  committed
  fd46244
- [Bugfix] Preserve soundfile-first automatic audio decoding
  tzhouam
  committed
  60b02c4
- [CI/Build] Remove completed L4 diagnostic lane
  tzhouam
  committed
  91ce937
- [Doc] Align configuration references and NPU guidance with vLLM 0.32
  tzhouam
  committed
  80e4791
- [Doc] Clarify release Python range and CUDA dependency ownership
  tzhouam
  committed
  a7e8131
- [CI/Build] Make Drama3 streaming quality checks reproducible
  tzhouam
  committed
  737c90c
Commits on Oct 19, 2026

Commits

Commits on Oct 1, 2026

Commits on Oct 2, 2026

Commits on Oct 3, 2026

Commits on Oct 4, 2026

Commits on Oct 5, 2026

Commits on Oct 18, 2026

Commits on Oct 19, 2026