Michael Munn

Learning without training: The implicit dynamics of in-context learning

On residual network depth

A Bayesian Model Selection Criterion for Selecting Pretraining Checkpoints

Leveraging free energy in pretraining model selection for improved fine-tuning