← Back to Test-Time Training
score-first TTT-like n-gram cache
Test-Time TrainingUsed in
1 PRs
Best BPB
0.9393
Avg BPB
0.9393
Submissions
Hyperparameters Across PRs
| pr_number | parameters |
|---|---|
| 810 | {"cache_updated_after_scoring":true,"per_gpu_independent_cache":true} |