← Back to Weight Averaging

Polyak averaging

Weight Averaging
Used in
7 PRs
Best BPB
0.2071
Avg BPB
0.7113

Hyperparameters Across PRs

pr_numberparameters
700{"decay":0.998}
731{"decay":0.998}
803{"decay":0.998}
809{"decay":0.998}
850{"decay":0.998}
851{"decay":0.998}
1014