Views
No views yet
stats/{Mean,Std,ActiveDims,ConstFill}.npy — dataset stats used to
normalize 310-dim velocity-space features at inference time.vqvae/net_best_fid.pth — frozen VQ-VAE tokenizer (codebook 512x512,
4x temporal downsample). Required for both encoding and decoding.gpt/net_last.pth — final iter-100k checkpoint of the unconstrained GPT.
(net_best_fid.pth is iter ~5k due to a CE-gating quirk and is not
the trained model — net_last.pth is the one to use.)| GPT layers / dim / heads | 4 / 512 / 8 |
| FFN multiplier | 4 |
| Block size (tokens) | 51 (~204 frames @ 30 fps after 4x downsample) |
| VQ-VAE codebook | 512 codes x 512 dim |
| Feature dim | 310 (pruned velocity-space SOMA-30) |
1from huggingface_hub import snapshot_download
2local = snapshot_download("mpilligua/car-t2m-small-unconstrained")
3# Then load like any local checkpoint dir; see CAR-T2M demo/inference.py.refactor).