← Back to Architecture
Soft MoE
ArchitectureUsed in
1 PRs
Best BPB
1.1826
Avg BPB
1.1826
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 660 | {"num_experts":2,"moe_layers":"last 2 layers","moe_start_layer":8} |