Skip to main content
Search

# Split-and-scale

1 related articles

Subscribe to updates
  1. arXiv

    3D latents stop paying for detail in tokens: per-voxel normalization lets one tokenizer scale at inference

    Latent 3D representations are capped by latent capacity, so complex objects lose fine detail, while scaling capacity at training time runs into memory limits. Split-and-scale divides a scene into sub-regions, normalizes each back to the original coordinate range, and runs the same pretrained tokenizer on each, so equal capacity covers a smaller extent and detail rises at inference. On reconstruction it matches LiTo models trained with the same total latent count and beats naive inference-time token scaling; its autoregressive generator is competitive with existing baselines.

You’re all caught up