← Back to Architecture
recursive weight sharing
ArchitectureUsed in
1 PRs
Best BPB
1.1355
Avg BPB
1.1355
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 579 | {"unique_blocks":6,"loops":2,"effective_depth":12,"MLP_expansion":"4x"} |