← Back to Regularization
LN scaling
RegularizationUsed in
1 PRs
Best BPB
1.0465
Avg BPB
1.0465
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 758 | {"scale":"1/sqrt(layer+1)"} |
| pr_number | parameters |
|---|---|
| 758 | {"scale":"1/sqrt(layer+1)"} |