← Back to Test-Time Training

causal TTT

Test-Time Training
Used in
2 PRs
Best BPB
1.1257
Avg BPB
1.1257

Submissions

PR #375by charmquark1984
1.1257
PR #375by charmquark1984
1.1257

Hyperparameters Across PRs

pr_numberparameters
375{"learning_rate":0.0001,"chunk_size":32000}
375{"learning_rate":0.01,"scope":"last 2 blocks MLP only"}