SLAT VQ tokenizers (512-code) — SS / shape / tex
Three frozen VQ tokenizers over TRELLIS.2 SLAT latents (Kyvo-style densify + 3D-conv + VQ),
512 codes/section. Used by weikaih/3d-vlm-stage1-v2.2. Load via torch.load(...) -> ck['args'], ck['model'],
ck['mean'], ck['std']. Code: https://github.com/weikaih04/3d_vlm
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support