PG Field Guide
Learn
Techniques
Emerging
PRs
← PR #1877
PR #1879→
PR #1878
open
val_bpb 0.85018 (3-seed mean) Raphe_II
by newjordan
View on GitHub
val_bpb
0.8502
Architecture
Transformer
Optimizer
—
Artifact Size
15,995,307 bytes
Training Techniques
Novel Contributions
Three-seed mean submission
Raphe_II model variant