Skip to content

vllm_omni.platforms.npu.models

Modules:

Name Description
cosyvoice2_dit_attn

NPU patches for CosyVoice2 / Token2Wav DiT attention.

minicpmo_4_5_code2wav

Inject MiniCPM-o Code2Wav NPUGraph acceleration on Ascend.

minimax_h3

NPU patches for the MiniMax H3 Qwen3-VL text encoder.

moss_tts_streaming_decode_wrapper

MOSS-TTS codec streaming NPUGraph acceleration on Ascend.

qwen3_tts

Patch Qwen3-TTS NPU runtime setup and weights.

qwen3_tts_tokenizer_v2

NPU patches for the Qwen3-TTS 12Hz tokenizer decoder.

step_audio2_token2wav

NPU patches for Step-Audio2 / MiniCPM Token2Wav.

voxcpm2_talker

Enable VoxCPM2 LocDiT NPUGraph acceleration on Ascend.