← Back to Architecture
KV heads
ArchitectureUsed in
1 PRs
Best BPB
1.1634
Avg BPB
1.1634
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 552 | {"kv_heads":4,"attention_heads":8} |
| pr_number | parameters |
|---|---|
| 552 | {"kv_heads":4,"attention_heads":8} |