Isotr0py
3affc7766d
* init Signed-off-by: Isotr0py <mozf@inferact.ai> * fix Signed-off-by: Isotr0py <mozf@inferact.ai> * test(diffusion): add Qwen-Image-2.1 unit tests Signed-off-by: Isotr0py <mozf@inferact.ai> * fix(diffusion): Qwen-Image-2.1 VAE spatial ratio, tiling e2e verification - AutoencoderKLQwenImage21: derive spatial_compression_ratio from the architecture (2 ** len(temperal_downsample) = 16x) instead of trusting the checkpoint's scale_factor_spatial=8, which corrupted tiled decode/encode geometry (output cropped to half size) and tiling gating. - pipeline: _read_vae_scale_factor prefers the architecture-derived value for the same reason. - models/qwen_image_21/__init__: lazy exports to break the circular import distributed VAE -> package __init__ -> pipeline -> distributed VAE. - Add VAE unit tests: ratio derivation, tiled vs untiled decode/encode consistency, distributed tile_split/exec/merge == tiled_decode. - e2e matrix (GB200, seed 42, 20 steps) all green; see e2e_out/qwen21_vae/. - examples: qwen_image_21_edit.py gains --vae-use-tiling. - docs: mark VAE patch-parallel verified for Qwen-Image-2.1; recipe notes the default-tile banding limitation at 2048px+. * clean tests Signed-off-by: Isotr0py <mozf@inferact.ai> * fix cudnn determination Signed-off-by: Isotr0py <mozf@inferact.ai> --------- Signed-off-by: Isotr0py <mozf@inferact.ai>
hide
| hide | ||
|---|---|---|
|
Welcome to vLLM-Omni
Easy, fast, and cheap omni-modality model serving for everyone
<script async defer src="https://buttons.github.io/buttons.js"></script> Star Watch Fork
About
vLLM was originally designed to support large language models for text-based autoregressive generation tasks. vLLM-Omni is a framework that extends its support for omni-modality model inference and serving:
- Omni-modality: Text, image, audio, video, and action data processing
- Non-autoregressive Architectures: extend the AR support of vLLM to Diffusion Transformers (DiT) and other parallel generation models
- Heterogeneous outputs: from traditional text generation to multimodal and action outputs
vLLM-Omni is fast with:
- State-of-the-art AR support by leveraging efficient KV cache management from vLLM
- Pipelined stage execution overlapping for high throughput performance
- Fully disaggregation based on OmniConnector and dynamic resource allocation across stages
vLLM-Omni is flexible and easy to use with:
- Heterogeneous pipeline abstraction to manage complex model workflows
- Seamless integration with popular Hugging Face models
- Tensor, pipeline, data and expert parallelism support for distributed inference
- Streaming outputs
- OpenAI-compatible API server
- Full-duplex realtime serving with streaming audio input and output
vLLM-Omni seamlessly supports most popular open-source models on HuggingFace, including:
- Omni-modality models (e.g. Qwen3-Omni, MiniCPM-o 4.5, Cosmos3, HunyuanImage, BAGEL)
- TTS models (e.g. Qwen3-TTS, IndexTTS 2.5, dots.tts, CosyVoice3)
- Diffusion models — image, video, and audio generation (e.g. MiniMax H3, LTX-2.5, SANA-Video, Wan2.2)
- Robot-policy and action models (e.g. π0, GR00T-N1.7, DreamZero-DROID, InternVLA-A1)
For more information, checkout the following: