← Back to Regularization

L2 loss

Regularization
Used in
1 PRs
Best BPB
1.2067
Avg BPB
1.2067

Submissions

PR #1433by mtybadger
1.2067

Hyperparameters Across PRs

pr_numberparameters
1433{"applied_to":"approximate codebook quantization","schedule":"turned on at end of training"}