realnvp_imagenet64(pretrained: bool = False, overrides: object = {})Construct the RealNVP flow for the Imagenet 64 x 64 setup.
Paper-faithful configuration from Dinh et al., 2016 §4.1: five scales
(64 → 32 → 16 → 8 → 4) with coupling networks of 2 residual
blocks, the depth the paper drops to at this resolution.
Model Size
Parameters
pretrainedbool= FalseReserved for future weight registration; currently a no-op.
**overridesobject= {}Optional
RealNVPConfig field overrides forwarded into
the underlying config.Returns
RealNVPModelBare bijection configured with the Imagenet 64 x 64 setup.
Notes
Reference: Dinh, Sohl-Dickstein, and Bengio, "Density Estimation Using Real NVP", ICLR, 2017 (arXiv:1605.08803), §4.1 and Table 1. Reported validation result: 3.98 bits/dim (3.75 train).
Examples
>>> from lucid.models.generative.realnvp import realnvp_imagenet64
>>> model = realnvp_imagenet64().eval()
>>> model.input_dim
12288