Modules:
| Name | Description |
|---|---|
full_sample_graph | Dense decode sampler graphs captured and parity-checked during startup. |
higgs_audio_v3_code2wav | Stage 1 (codec decoder) for higgs-audio v3. |
higgs_audio_v3_talker | Stage-0 talker for higgs-audio v3 (Qwen3 backbone, fused multi-codebook). |
higgs_audio_v3_tokenizer | Prompt builder for higgs-audio v3 TTS. |
model_state | Higgs delay-pattern sampling and owned audio output on Model Runner V2. |
mrv2_state_kernels | Fused movement of model-owned delay state when a request batch changes. |
pipeline | higgs-audio v3 pipeline: Talker (text -> 8-codebook codec) -> Code2Wav (codec -> 24 kHz PCM). |
step_kernels | Fused per-step Higgs kernels for the MRV2 decode loop. |