vllm_omni.model_executor.models.yue2.pipeline ¶

YuE2-3B pipeline topology.

Single-stage native-AR text-to-music (gepard pattern). The whole Mixture-of-Transformers checkpoint — AR path, NAR path, projection heads — lives in one model.safetensors, so both the autoregressive sampling and the terminal flow-matching + VAE pass run inside one stage and no weights are duplicated across stages. The VAE decoder is a separate repository (m-a-p/YuE2-Vae), loaded in load_weights from $YUE2_VAE or the default hub id.

Sampling is model-owned. sampling_constraints pins detokenize and the union of both phase end tokens; extra_args carry the phase and the request-local sampling preset, and must be present in the stage's default_sampling_params so has_sampling_extra_args turns on and the per-request args reach the model's forward.

YUE2_PIPELINE = PipelineConfig(
    model_type="yue2",
    default_deploy_config_name="yue2.yaml",
    model_arch="Yue2ForCausalLM",
    stages=(
        StagePipelineConfig(
            stage_id=0,
            model_stage="yue2",
            execution_type=StageExecutionType.LLM_AR,
            input_sources=(),
            final_output=True,
            final_output_type="audio",
            owns_tokenizer=True,
            engine_output_type="audio",
            sampling_constraints={
                "detokenize": False,
                "stop_token_ids": list(STOP_TOKEN_IDS),
            },
        ),
    ),
)