← Back to Architecture
SwiGLU FFN
ArchitectureUsed in
2 PRs
Best BPB
1.0672
Avg BPB
1.0926
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 462 | — |
| 505 | {"hidden":1792} |
| pr_number | parameters |
|---|---|
| 462 | — |
| 505 | {"hidden":1792} |