← Back to Architecture

Mixture of Softmax

Architecture
Used in
2 PRs
Best BPB
1.3921
Avg BPB
1.3926

Submissions

PR #266by User123331
1.3932
PR #1608by User123331
1.3921

Hyperparameters Across PRs

pr_numberparameters
266{"k":2,"rank":64}
1608{"k":2,"rank":64}