PointCut model weights

Compact copies of open model weights that the PointCut video editor downloads on a Mac.

File	What it is	Source
`musetalkV15/unet.fp16.safetensors`	MuseTalk 1.5 UNet in half precision (fp16), 1.7 GB instead of 3.4 GB. Converted from `musetalkV15/unet.pth` (sha256 `7ebf6c98c181e20838e4c0054e96e944ac60d5d692cc01db42839fe11b787007`) with `tensor.to(torch.float16)`; nothing else changed. This is exactly the precision the Mac's GPU runs.	TMElyralab/MuseTalk, MIT license (Tencent Music Entertainment), see `LICENSE-MuseTalk`

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support