MNN 4-bit build of Qwen3.8-27B for on-device inference.
🤝 Open to Collab
Yunfeng Wang
yunfengwang
webbrain-one-318852's profile picture
anshuman-reasoncore's profile picture
developerabu's profile picture·
AI & ML interests
Computer Vision/LLM/Audio
Recent Activity
liked a model 6 days ago
taobao-mnn/MNN-Pocket-0.3B-GGUF liked a model 7 days ago
pfnet/plamo-3-translate-0.8b liked a model 8 days ago
taobao-mnn/MNN-Pocket-0.3BOrganizations
MNN conversion of MiniMax-Music3 for on-device text-to-music on Apple Silicon.
MNN build of Higgs-TTS-3 (4B, Qwen3 backbone).
MNN assets for CPU zero-shot voice cloning with Qwen3-TTS-12Hz-0.6B-Base.
MNN CLI and models for Supertonic TTS.
ONNX export of IndexTTS for CPU inference. See the separate IndexTTS-2.5 collection for its MLX/ONNX/MNN builds.
MNN fp16 and int8 builds of Laya, the non-autoregressive System 1 decision model.
ONNX (fp32) and MNN (fp16 + w8) builds of FireRedTTS3 for CPU text-to-speech inference.
MNN conversions of Qwen-Image-2.1 for on-device text-to-image inference (fp16 / int8 / int4).
Edge Models (MLX, ONNX, MNN) of IndexTTS-2.5
MLX weights of Anima-2.9B (DiT + Qwen3 text encoder + Qwen-Image VAE).
mmh3turbo bundles: prepared MLX weights for MiniMax-H3 video and audio generation on Apple Silicon.
MNN builds of Qwen3-ASR 0.6B and 1.7B for on-device speech recognition.
GGUF and companion weights for Wan2.2-TI2V-5B text/image-to-video generation.
MLX version of common-pile/comma-v0.1-1t.
On-device Z-Image-Turbo weights: MNN conversion and the zimgturbo int8 bundle.
MNN 8-bit and 4-bit builds of MiniCPM5-2B for on-device inference.
ONNX and MNN conversions of WavLM-Large for on-device audio feature extraction.
MNN 4-bit build of Qwen3.8-27B for on-device inference.
MLX weights of Anima-2.9B (DiT + Qwen3 text encoder + Qwen-Image VAE).
MNN conversion of MiniMax-Music3 for on-device text-to-music on Apple Silicon.
mmh3turbo bundles: prepared MLX weights for MiniMax-H3 video and audio generation on Apple Silicon.
MNN build of Higgs-TTS-3 (4B, Qwen3 backbone).
MNN builds of Qwen3-ASR 0.6B and 1.7B for on-device speech recognition.
MNN assets for CPU zero-shot voice cloning with Qwen3-TTS-12Hz-0.6B-Base.
GGUF and companion weights for Wan2.2-TI2V-5B text/image-to-video generation.
MNN CLI and models for Supertonic TTS.
MLX version of common-pile/comma-v0.1-1t.
ONNX export of IndexTTS for CPU inference. See the separate IndexTTS-2.5 collection for its MLX/ONNX/MNN builds.
On-device Z-Image-Turbo weights: MNN conversion and the zimgturbo int8 bundle.
MNN fp16 and int8 builds of Laya, the non-autoregressive System 1 decision model.
MNN 8-bit and 4-bit builds of MiniCPM5-2B for on-device inference.
ONNX (fp32) and MNN (fp16 + w8) builds of FireRedTTS3 for CPU text-to-speech inference.
ONNX and MNN conversions of WavLM-Large for on-device audio feature extraction.
MNN conversions of Qwen-Image-2.1 for on-device text-to-image inference (fp16 / int8 / int4).
Edge Models (MLX, ONNX, MNN) of IndexTTS-2.5