Neural Networks Learn Statistics of Increasing Complexity
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:22FWHIAWrecord.jsonopen to challenge →
read the original abstract
The distributional simplicity bias (DSB) posits that neural networks learn low-order moments of the data distribution first, before moving on to higher-order correlations. In this work, we present compelling new evidence for the DSB by showing that networks automatically learn to perform well on maximum-entropy distributions whose low-order statistics match those of the training set early in training, then lose this ability later. We also extend the DSB to discrete domains by proving an equivalence between token $n$-gram frequencies and the moments of embedding vectors, and by finding empirical evidence for the bias in LLMs. Finally we use optimal transport methods to surgically edit the low-order statistics of one class to match those of another, and show that early-training networks treat the edited samples as if they were drawn from the target class. Code is available at https://github.com/EleutherAI/features-across-time.
This paper has not been read by Pith yet.
Forward citations
Cited by 3 Pith papers
-
When Are Two Networks the Same? Tensor Similarity for Mechanistic Interpretability
Tensor similarity is a symmetry-invariant metric that measures functional equivalence between tensor-based networks using a recursive algorithm for cross-layer mechanisms.
-
Distributional simplicity bias and effective convexity in Energy Based Models
Gradient flow in energy-based models for strictly positive binary distributions produces stable data-consistent fixed points and a learning hierarchy that favors lower-order interactions first, mechanistically explain...
-
A theory of learning data statistics in diffusion models, from easy to hard
Diffusion models exhibit a distributional simplicity bias, learning pairwise input statistics at linear sample complexity while fourth-order cumulants require cubic complexity unless sharing correlated latent structure.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.