Diffusion model with a Transformer backbone + SPRINT sparse-dense residual fusion.

r"""Collects the scalars broadcasted by `report()` and `report0()` and computes their long-term averages (mean and standard deviation) over us

Calculate pairwise distances between vectors. Adapted from https://github.com/kynkaat/improved-precision-and-recall-metric/blob/f60f25e5ad93

The final layer of SiT.

Embeds class labels into vector representations. Also handles label dropout for classifier-free guidance.

A helper for comparing manifolds of feature vectors. Adapted from https://github.com/kynkaat/improved-precision-and-recall-metric/blob/f60f2

A SiT block with adaptive layer norm zero (adaLN-Zero) conditioning.

Embeds scalar timesteps into vector representations.

Fast EVA-02 2D RoPE with broadcasting, extended to support: • per-token rope_ids (routing / unsorted subsets) • leading CLS tokens (u

Convenience class that behaves like a dict but allows access with the attribute syntax.

Redirect stderr to stdout, optionally print stdout to a file, and optionally force flushing on both stdout and the file.