We leave the `dataset.json` file in the same format as in the original SG2-ADA repo: it's `labels` field is a hashmap of filename-lab

r"""Hook that is called internally by the `pickle` module to unpickle a persistent object.

r"""Returns the averages accumulated between the last two calls to `update()` as an `dnnlib.EasyDict`. The contents are as follows:

Ask the user the question until the user inputs a valid answer.

Create a 1D, 2D, or 3D average pooling module.

r"""Fused bias and activation function. Adds bias `b` to activation tensor `x`, evaluates activation function `act`, and scales the result by

Compute the entire variational lower-bound, measured in bits-per-dim, as well as other related quantities. :param model: t

Center cropping implementation from ADM. https://github.com/openai/guided-diffusion/blob/8fb3ad9197f16bbc40620447b2742e13458d2831/guided_di

Evaluate a function without caching intermediate activations, allowing for reduced memory at the expense of extra compute in the backward p