← Back to Optimizer

MUD

Optimizer
Used in
1 PRs
Best BPB
1.1989
Avg BPB
1.1989

Hyperparameters Across PRs

pr_numberweight_decaymomentumother_params
510{"mud_whiten_replaces":"zeropower_via_newtonschulz5","passes":1,"eps":1e-7}