MaRN

MaRN: a PyTorch library for low-dimensional parameter mapping

The author released MaRN, a PyTorch library that optimizes a compact latent representation instead of all model parameters. Benchmarks show an MNIST CNN cut from 107,998 to 1,872 trainable parameters (57.7x) at 91.80% accuracy, though mapped models train much slower; the author calls results exploratory, not evidence of general superiority.