Unlocking Tuning-free Generalization: Minimizing the PAC-Bayes Bound with Trainable Priors

Open in new window