anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity
2,905 Commits 235 Branches 31 Tags
285db2fa3ce4d8a215bd1880af863e5aa99bdd6d
Commit Graph
2000 Commits
Author SHA1 Message Date
tzhouamandClaude Fable 5.1 8e912db6c8 [Edge][WP6] Slim the AR/TTS import path and add an edge dependency set 
* diffusers is no longer imported by vllm_omni.diffusion.data (annotation only);
  vllm_omni.diffusion.io_support (diffusion registry + distributed state) is
  imported lazily by config_factory and async_omni_engine, so constructing a
  Qwen3-TTS pipeline config / AsyncOmni no longer loads diffusers,
  x_transformers or whisper (guarded by tests/config/test_import_guard.py).
* requirements/common.txt = core.txt (AR/speech runtime) + media.txt
  (diffusion/video/extra audio); VLLM_OMNI_EDGE_BUILD=1 makes setup.py use
  requirements/<device>-edge.txt (core only); `edge` extra adds websockets and
  sounddevice for on-device clients.

Measured on this host (CUDA venv): `import vllm` 7.0-7.9 s, `import vllm_omni`
adds ~20 s of which ~17 s is transformers' own import (transformers.utils.import_utils)
pulled in by vllm_omni.patch / vllm_omni.version; that floor is not
vLLM-Omni's and is unchanged. Guard test: no diffusion module loaded on the TTS path.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 11:14:50 +08:00
tzhouamandClaude Fable 5.1 49a839ecf9 [Edge][WP5] Add codec-token streaming mode with on-device decoding 
* qwen3_tts_talker_only pipeline (single terminal LLM_AR stage,
  final_output_type "latent") + vllm_omni/deploy/qwen3_tts_talker_only.yaml;
  a terminal talker publishes its per-step codes.audio rows to the client
  (omni_client_multimodal_output_keys).
* /v1/audio/speech/stream: session.config.output_mode="codec_tokens" streams
  codec.start (JSON) + binary frames (<IHB seq/n_frames/flags + int16 rows,
  vllm_omni/entrypoints/openai/codec_stream.py) + codec.done; new input.cancel
  message aborts an in-flight generation for both codec and PCM sessions
  (single reader task + session inbox replaces direct socket reads); optional
  session.config.seed; audio.done gains "cancelled" only when it happened.
* serving_speech._generate_codec_chunks aligns the cumulative code tensor to
  output token ids (drops prefill placeholder/EOS rows) and emits deltas.
* TTS adapters: supports_codec_stream / codec_stream_spec (Qwen3-TTS: 16
  codebooks, 2048 ids, 12.5 Hz, 24 kHz, decoder qwen3_tts_tokenizer_v2_12hz).
* examples/online_serving/text_to_speech/qwen3_tts/edge_decoder_client.py
  decodes locally with Qwen3TTSTokenizerV2Decoder (ramp schedule, 72-frame
  context, cancel/reset), codec_parity_check.py and
  benchmarks/tts/codec_stream_e2e.py compare against the two-stage server.
* docs/serving/codec_token_streaming.md; CPU unit tests for the wire format,
  alignment, session config and handler (frames, unsupported adapter, cancel);
  L3 e2e tests/e2e/online_serving/test_qwen3_tts_codec_stream.py.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 11:10:13 +08:00
tzhouamandClaude Fable 5.1 35e43b9983 [Edge][WP3] Add CPU OmniPlatform (AR + generation stages on vLLM's CPU backend) 
* vllm_omni/platforms/cpu/platform.py: CPUOmniPlatform(OmniPlatform, CpuPlatform)
  selected when VLLM_TARGET_DEVICE=cpu or a +cpu vLLM wheel is installed
  (builtin plugin "cpu"; explicit CPU target short-circuits accelerator probes
  like vLLM does). NUMA-based memory queries, one logical device, no graph
  wrapper (sentinel type for isinstance checks), inductor opt-in via
  VLLM_OMNI_CPU_INDUCTOR=1, diffusion stages unsupported.
* vllm_omni/platforms/cpu/worker: CPUARWorker / CPUGenerationWorker compose
  vLLM's CPUWorker with the Omni mixin (XPU pattern); CPUARModelRunner /
  CPUGenerationModelRunner compose the Omni runners with CPUModelRunner
  (no graphs, no async omni output; sleep/wake return error ACKs).
* patch.py skips the CuMem patch on CPU-only torch; Qwen3-TTS speaker encoder
  and prompt-embeds dtype follow the model dtype (float32/float16 on CPU).
* vllm_omni/deploy/qwen3_tts.yaml: `platforms: cpu:` overlay (eager, no async
  scheduling, absolute kv_cache_memory_bytes, disjoint OMP thread ranges).
* requirements/cpu.txt (+psutil, intel-openmp), docs/getting_started/installation/cpu.md.

Measured (this x86 Xeon 8480C host, vllm 0.28.0+cpu, bf16, 12+4 threads):
Qwen3-TTS 0.6B CustomVoice end-to-end on CPU, 12 prompts: TTFA p50 292 ms,
RTF ~1.7 (25-frame chunk ~4.6 s), init 185 s. Result JSON + init timeline
committed under benchmarks/tts/edge_results/.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 11:02:14 +08:00
tzhouamandClaude Fable 5.1 55192de984 [Edge][WP4] Add hardware adaptation: probe, derived overrides, auto profile, calibration 
* vllm_omni/edge/hardware_probe.py: HardwareProfile from /proc, /sys and torch
  (ISA flags incl. AVX-512/AMX and NEON/dotprod/i8mm/SVE, physical cores,
  big.LITTLE clusters, NUMA, RAM, CUDA discrete vs Tegra unified memory, NPU
  via override), hardware classes jetson / cuda_discrete / arm64_cpu / x86_cpu /
  npu_phone; VLLM_OMNI_HW_PROFILE=<json> overrides the probe (emulation, tests).
* vllm_omni/edge/adapt.py: derive_overrides -> dtype, absolute KV budgets from
  available memory minus weights, max_num_seqs, per-stage disjoint CPU thread
  ranges on the big cluster, eager/graph switches for small or unified memory,
  adaptive-chunk parameters, timeouts from a load cost model.
* vllm_omni/edge/calibrate.py: chunk-schedule calibration from measured chunk
  arrivals with a per-hardware-fingerprint cache under VLLM_CACHE_ROOT.
* vllm_omni/edge/auto_profile.py + deploy_profile="auto": edge/<model>.yaml ->
  edge/hardware/<class>.yaml -> derived overrides -> cached calibration,
  materialized as a temp deploy YAML (platform overlays are superseded).
* benchmarks/tts/hw_emulation.py: emulated classes (ATEN_CPU_CAPABILITY,
  taskset, profile overrides) driving stream_latency_bench with auto profiles.
* CPU-only tests for probe snapshots, derivation rules, calibration and auto.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 11:01:40 +08:00
tzhouamandClaude Fable 5.1 073d31b9cd [Edge][WP2] Add initialization timeline and cold-start profiler 
* vllm_omni/engine/init_timeline.py: InitTimeline phase collector; engine
  phases (stages_init, g3_device_lock, g2_spawn/ready) and worker phases
  (proc_entry, engine_core_init, init_device, load_model/load_weights,
  talker_mtp_init, attn_prewarm, cudagraph_capture, talker_mtp_capture,
  code2wav_cudagraph, predictor_setup_compile/warmup/cudagraph) recorded via
  worker_phase/timed_phase; workers append JSONL under
  VLLM_OMNI_INIT_TIMELINE_DIR (inherited at spawn) and the engine merges them.
  Enabled by VLLM_OMNI_INIT_TIMELINE=<json> or log_stats; table logged at INFO,
  partial table on orchestrator startup timeout, suggested timeouts logged.
* omni_base.py: named deploy profiles default to stage_init_timeout=900 /
  init_timeout=1500 unless given explicitly.
* gpu_model_runner._init_talker_mtp tolerates max_cudagraph_capture_size=None;
  gpu_ar_model_runner.shutdown skips CUDA graph teardown on CPU.
* benchmarks/tts/init_profile.py: runs stream_latency_bench --init-only per
  configuration (default, cold_cache, eager_stage1, eager_both,
  parallel_stage_init, edge, ...) and tabulates phase durations.
* stream_latency_bench: --parallel-stage-init / --deploy-profile flags; the
  resource poller no longer shadows Thread._stop and skips GPU sampling on CPU.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 11:01:39 +08:00
tzhouamandClaude Fable 5.1 b22dcc59f7 [Edge][WP1] Add edge deploy profile and deploy_profile selector 
* vllm_omni/deploy/edge/qwen3_tts.yaml: single-device Qwen3-TTS profile with a
  ramped codec chunk schedule (2->4->8->16->25 frames) that removes the ~80 ms
  playback stall of the default 1->25 schedule, decode batch 1, max_num_seqs 4.
* deploy_profile="edge" (Omni/AsyncOmni kwarg, --deploy-profile CLI, headless
  serve) resolves vllm_omni/deploy/<profile>/<model_type>.yaml; mutually
  exclusive with deploy_config; explicit profiles never fall back silently.
* tests/helpers/stage_config.py: modify_stage_config anchors a relative
  base_config before writing the temp YAML so overlay-derived configs resolve.
* async_omni_engine.py also carries the init-timeline hooks used by WP2.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-09-09 10:27:33 +08:00
Lei KeandKrystalRay 7be014bce6 [Bugfix][NPU] Fix MiniMax-H3 INT8 quantization dispatch (#6876) 
Signed-off-by: KrystalRay <keeleiray@gmail.com>
Co-authored-by: KrystalRay <keeleiray@gmail.com>
 2026-09-07 14:50:44 +00:00
汪志鹏 9b6a308914 [Perf][Diffusion] Keep LoRA weights resident across activation cycles (#7195) 
Signed-off-by: princepride <wangzhipeng628@gmail.com>
 2026-09-07 14:27:24 +00:00
WeiQing Chen 30d6a0b4e6 [Bugfix][MiniMax-H3] Stabilize keyframe VAE encoding (#7191) 
Signed-off-by: david6666666 <530634352@qq.com>
Signed-off-by: David Chen <530634352@qq.com>
 2026-09-07 21:52:54 +08:00
Mu GuanLin 3204a0b2ad [Model][Perf] Optimize LTX Ulysses sequence parallelism (#7079) 
Signed-off-by: mglyn <1203789601@qq.com>
 2026-09-07 10:48:26 +00:00
Rahul SteigerandRahul Steiger f080f41e03 [Bugfix][Cosmos3] Auto-pad non-divisible Ulysses GEN sequences (#6918) 
Signed-off-by: Rahul Steiger <rsteiger@nvl72d188-T17.cm.cluster>
Co-authored-by: Rahul Steiger <rsteiger@nvl72d188-T17.cm.cluster>
 2026-09-07 08:58:28 +00:00
FredHuangNV 85b0bad2d3 [Frontend][Model] Add Cosmos3 control uploads to video API (#7027) 
Signed-off-by: Fred Huang <frhuang@nvidia.com>
 2026-09-07 01:38:56 -07:00
Rahul SteigerRahul SteigerRahul Steiger
 		677265b64b [Bugfix][Diffusion] Make Cosmos3 transfer image arrays writable (#6915) 
Signed-off-by: Rahul Steiger <rsteiger@nvl72d179-T17.cm.cluster>
Co-authored-by: Rahul Steiger <rsteiger@nvl72d179-T17.cm.cluster>
Co-authored-by: Rahul Steiger <rsteiger@aws-cmh-slurm-1-vscode-04.cm.cluster>
 2026-09-07 01:06:16 -07:00
Shenglei FuandClaude 91dfafcaab [Bugfix][Diffusion] Include guidance_scale_2_provided in the request-batch key (#7078) 
Signed-off-by: Shenglei Fu <sfu@confluent.io>
Co-authored-by: Claude <noreply@anthropic.com>
 2026-09-07 15:58:19 +08:00
summerandAlicia 4b81f692d1 [Bugfix][Diffusion] Canonicalize cache_backend=None to "none" (#7041) 
Signed-off-by: summer <128961079+zhang-keliang@users.noreply.github.com>
Co-authored-by: Alicia <115451386+congw729@users.noreply.github.com>
 2026-09-07 15:56:51 +08:00
BruceLoveDecimal 5cd925631c [Feature][LingBot World]stepwise execution (#6844) 
Signed-off-by: BruceLoveDecimal <liuqihao970610@gmail.com>
Signed-off-by: liuqihao <liuqihao970610@gmail.com>
 2026-09-07 13:38:23 +08:00
NumberWanandAlicia 7f32c326c6 [BugFix][Qwen-Image-Edit] restore txt_seq_lens RoPE width after #3588 regression (#5586) 
Signed-off-by: NumberWan <wantszkin2003@gmail.com>
Co-authored-by: Alicia <115451386+congw729@users.noreply.github.com>
 2026-09-07 12:10:47 +08:00
linzhenpl07andClaude Sonnet 5 38abc0ba28 [AR-Diffusion] Allocate K and V separately so the compiled path stops cloning the pool (#6463) 
Signed-off-by: linzhenpl07 <linzhenpl07@gmail.com>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
 2026-09-07 11:40:20 +08:00
Zeyu Huang | 黃澤宇 2df95de421 [Core][Diffusion] Refactor prompt updates into a modality-ready interaction framework (#6294) 
Signed-off-by: Huang, Zeyu <11222265+fhfuih@users.noreply.github.com>
 2026-09-07 11:40:04 +08:00
汪志鹏 eeee257913 [Bugfix][Diffusion] Restore the TRTLLM attention default for the MiniMax-H3 modular alias (#7162) 
Signed-off-by: princepride <wangzhipeng628@gmail.com>
 2026-09-07 07:21:21 +08:00
NATUREandamy-why-3459 ff6e906a25 [Model] Preserve MiniCPM-o Talker full attention until capacity (#6799) 
Signed-off-by: natureofnature <wzliu@connect.hku.hk>
Co-authored-by: amy-why-3459 <wuhaiyan17@huawei.com>
 2026-09-06 22:25:35 +08:00
Hongsheng Liu 715b8b8746 [Model] MiniMax-H3: validate versioned text-conditioning handoff (#6720) 
Signed-off-by: hsliuustc0106 <liuhongsheng4@huawei.com>
 2026-09-06 21:45:10 +08:00
SYLARandhsliuustc0106 14d2d00536 [Bugfix] Fix Hunyuan Image-3 expert mapping contract (#7021) 
Signed-off-by: hsliuustc0106 <liuhongsheng4@huawei.com>
Co-authored-by: hsliuustc0106 <liuhongsheng4@huawei.com>
 2026-09-06 21:43:37 +08:00
汪志鹏andClaude Opus 5 af74a5a155 [Feature][MiniMax-H3] Support the full LightX2V Turbo LoRA matrix (#7062) 
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
 2026-09-06 20:23:36 +08:00
XenorynandClaude 039808e0d9 [Perf][Diffusion] Skip the Ulysses length all-gather when shards are padded equal (#5717) 
Signed-off-by: Xenoryn <timberh2000@outlook.com>
Co-authored-by: Claude <noreply@anthropic.com>
 2026-09-06 16:23:30 +08:00
DMAN 06541d1f52 [Model] FP8 Support For Boogu-Image (#6925) 
Signed-off-by: Dmaner <2663515256@qq.com>
 2026-09-06 04:48:38 +00:00
prettygirlisnotme 32370968d0 [Model] Add Boogu-Image Turbo native DMD path (#6699) 
Signed-off-by: Dragonborn <1078757180@qq.com>
 2026-09-06 08:44:15 +08:00
Zeyu Huang | 黃澤宇asukaqaq-sHongsheng Liu
 		f26f9c3a7f [Diffusion] Implement paged AR↔DiT KV connector ("v1", reusing vllm's native mooncake) (#6310) 
Signed-off-by: asukaqaq-s <1311722138@qq.com>
Signed-off-by: Huang, Zeyu <11222265+fhfuih@users.noreply.github.com>
Co-authored-by: asukaqaq-s <1311722138@qq.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
 2026-09-06 05:41:54 +08:00
SYLAR ea3cf99894 [diffusion][feature] Add component-selective offload policies (#5929) 
Signed-off-by: lishunyang12 <lishunyang12@163.com>
 2026-09-06 05:40:09 +08:00
Samit 44d3ae100a [Bugfix][Diffusion] Fuse distilled LoRA weights before HSDP sharding (#6948) 
Signed-off-by: samithuang <285365963@qq.com>
Signed-off-by: Samit <285365963@qq.com>
 2026-09-06 01:29:59 +08:00
Zhou TaichangandClaude Fable 5 142151fef5 Parallel stage initialization (admission + SH/EX device locks) (Worker refactor 1/N) (#5224) 
Signed-off-by: Taichang Zhou <tzhouam@connect.ust.hk>
Signed-off-by: tzhouam <tzhouam@connect.ust.hk>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
 2026-09-05 22:22:41 +08:00
chickeytonClaude Fable 5Gao Han
 		5d7fdf9350 [Core] Graduate MiniCPM-o 4.5 and PersonaPlex full-duplex serving out of experimental (#6196) 
Signed-off-by: chickeyton <ngton2014@gmail.com>
Signed-off-by: Gao Han <hgaoaf@connect.ust.hk>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: Gao Han <hgaoaf@connect.ust.hk>
 2026-09-05 08:20:21 +00:00
Hongsheng Liu d4096e01fe [Model] Add native MAGI-2 Preview diffusion support (#5918) 
Signed-off-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Signed-off-by: hsliuustc0106 <liuhongsheng4@huawei.com>
 2026-09-05 15:32:30 +08:00
vOv 13c617a140 [Model] Add sequence parallelism to SANA-Video 2B (#5940) 
Signed-off-by: cr-gao <gaochenrui@sjtu.edu.cn>
 2026-09-05 15:22:53 +08:00
boatman 66bbebd9d3 [Bugfix] Fix fallback loading for IndexTTS external models (#6137) 
Signed-off-by: boatman <109857087+sphinxkkkbc@users.noreply.github.com>
 2026-09-04 19:58:33 -07:00
BruceLoveDecimalandClaude Opus 5 1e74807c35 [Feature][Moss-TTS] Reference Audio Preprocessing (#4982) 
Signed-off-by: BruceLoveDecimal <39156883+BruceLoveDecimal@users.noreply.github.com>
Signed-off-by: liuqihao <liuqihao970610@gmail.com>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
 2026-09-04 16:36:26 -07:00
LOGO127 b6b992ac7b [Bugfix] Fix Higgs Audio v3 voice-clone token validation (#7065) 
Signed-off-by: luozijian <luozijian0924@gmail.com>
 2026-09-04 16:06:47 -07:00
Inesh Reddy ChappidiandGao Han 66d938100b [Bugfix] Give exact-shape codec models an unpadded input_ids view (#6775) 
Signed-off-by: Inesh Reddy <ineshreddy249@gmail.com>
Signed-off-by: IneshReddy249 <ineshreddy249@gmail.com>
Co-authored-by: Gao Han <hgaoaf@connect.ust.hk>
 2026-09-04 15:28:49 -07:00
boatman 348fd0fc07 [Feat][dots.tts] Add Online Serving Support (#6235) 
Signed-off-by: boatman <109857087+sphinxkkkbc@users.noreply.github.com>
 2026-09-04 14:44:10 -07:00
TriggerandGao Han ea14b04e9f [Bugfix][BAGEL] Restore original denoise step count (#7049) 
Signed-off-by: Trigger <Meng.Bo.Wang@outlook.com>
Signed-off-by: Gao Han <hgaoaf@connect.ust.hk>
Co-authored-by: Gao Han <hgaoaf@connect.ust.hk>
 2026-09-04 17:05:05 +00:00
SYLARandhsliuustc0106 22cefe04b4 [Bugfix] Update Hub kernel dependency floor (#7020) 
Signed-off-by: hsliuustc0106 <liuhongsheng4@huawei.com>
Co-authored-by: hsliuustc0106 <liuhongsheng4@huawei.com>
 2026-09-04 22:15:55 +08:00
SYLARandhsliuustc0106 7493cbd3c2 [Bugfix] Fix Qwen3-Omni MoE backend selection (#7019) 
Signed-off-by: hsliuustc0106 <liuhongsheng4@huawei.com>
Co-authored-by: hsliuustc0106 <liuhongsheng4@huawei.com>
 2026-09-04 21:16:58 +08:00
MikeyDong1 41f4f9d8c9 [Perf][DreamZero] Remove redundant CUDA compile guard (#6983) 
Signed-off-by: MikeyDong1 <xdong-1219@outlook.com>
 2026-09-04 09:01:05 -04:00
heyuanliu-intel 10701e1838 [Bugfix] Release diffusion worker RPC results on ranks that do not reply (#6989) 
Signed-off-by: heyuanliu-intel <heyuan.liu@intel.com>
 2026-09-04 16:45:17 +08:00
Shenglei FuandClaude Opus 4.8 0e903e9695 [Diffusion] Enable request-level batching for Boogu-Image (T2I) (#6968) 
Signed-off-by: Shenglei Fu <117230642+ShengleiFu@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-09-04 16:27:51 +08:00
Mu GuanLin 175dec11be [Perf][Model] Optimize LTX video output transport (#7000) 
Signed-off-by: mglyn <1203789601@qq.com>
 2026-09-04 06:47:28 +00:00
Yuekai ZhangandClaude Fable 5 4179b0cae1 [Perf][NemotronVoiceChat] Fast defaults for all three deploy profiles: 2.8x offline, realtime full-duplex audio (#6354) 
Signed-off-by: Yuekai Zhang <zhangyuekai@foxmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
 2026-09-03 18:49:19 -07:00
Trigger e43c787544 [Model] Add BAGEL step execution and continuous batching (#6359) 2026-09-04 01:36:11 +00:00
Bo Li fda3af8bbc [Bugfix] Preserve BF16 residual semantics in fused MiniMax-H3 modulation (#6878) 
Signed-off-by: Bo Li <22713281+bobboli@users.noreply.github.com>
 2026-09-04 02:01:34 +08:00
Mu GuanLin 1dae537313 [Bugfix][LTX] Restore distilled two-stage video serving (#6847) 
Signed-off-by: mglyn <1203789601@qq.com>
 2026-09-03 18:48:43 +08:00