← Back to Architecture
MHA
ArchitectureUsed in
2 PRs
Best BPB
1.0618
Avg BPB
1.1298
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 1052 | {"kv_heads":8} |
| 1987 | {"num_kv_heads":8} |
| pr_number | parameters |
|---|---|
| 1052 | {"kv_heads":8} |
| 1987 | {"num_kv_heads":8} |