← Back to Architecture

FlashAttention-3

Architecture
Used in
3 PRs
Best BPB
0.6864
Avg BPB
1.0175

Submissions

PR #614by bigbag
0.6864
PR #640by CiprianFlorin-IfrimRECORD
1.1570
PR #1427by kjahan
1.2092

Hyperparameters Across PRs

pr_numberparameters
614—
640—
1427—