← Back to Architecture

VRL

Architecture
Used in
9 PRs
Best BPB
0.4416
Avg BPB
0.9883

Hyperparameters Across PRs

pr_numberparameters
175{"layers":11,"gate_init":-1.5,"initial_mixing":0.18}
457{"layers":[1,10]}
670
731
738
803
887
889
915