AVTR-1 MLX for AI2Apps
Portable MLX conversion of AVTR-1 with supporting Ditto rendering graphs. This is a third-party conversion, not an official Avaturn release. Original terms remain applicable; accepting or downloading this mirror does not grant additional rights. Read LICENSE-MODEL.md, LICENSE-RENDERER.md, PATENTS.md and ATTRIBUTION.md before use.
Original motion/audio source: avaturn-live/avtr-1 at 4c9bd5550f2617d0409ac602f75378502f675731. Renderer graphs: digital-avatar/ditto-talkinghead at e4a2f60328ee7c32af585ac4b3cce299e4c8e254. Full per-file provenance and converted checksums are in ai2apps-checkpoint.json. All model inference uses MLX; graph.json and safetensors replace ONNX at runtime. No Python pickles or TorchScript are needed for inference. Use the matching AI2Apps AVTR-1 model Package.
- Downloads last month
- 16
Quantized