PG Field Guide
Learn
Techniques
Emerging
PRs
← Back to Architecture
MHA
Architecture
Used in
1 PRs
Best BPB
1.1978
Avg BPB
1.1978
Submissions
PR #1052
by demouo
1.1978
Hyperparameters Across PRs
pr_number
parameters
1052
{"kv_heads":8}