← Back to Architecture
shared attention
ArchitectureUsed in
2 PRs
Best BPB
1.3527
Avg BPB
1.3553
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 1025 | {"layers":11,"bases":10,"rank":128} |
| 1235 | {"shared_pairs":4} |