← Back to Architecture
GQA / KV head count
ArchitectureUsed in
1 PRs
Best BPB
1.1472
Avg BPB
1.1472
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 179 | {"layers":11,"num_heads":8,"num_kv_heads":4} |