List your company
ResearchFormerly WizWand

Language Modeling on Political debates (held-out)

2.61PerplexityRooseBERT
-3.462437.526378.515119.5037Aug 5, 2025
Updated 8mo ago

Evaluation results

MethodLinks
RooseBERTCase sensitivity=cased, Pre-training strategy=cont2025.082.61
RooseBERTCase sensitivity=uncased, Pre-training strategy=cont2025.082.71
RooseBERTCase sensitivity=cased, Pre-training strategy=scr2025.082.8
RooseBERTCase sensitivity=uncased, Pre-training strategy=scr2025.083.09
ConfliBERTCase sensitivity=cased, Pre-training strategy=cont2025.084.37
ConfliBERTCase sensitivity=cased, Pre-training strategy=scr2025.084.66
ConfliBERTCase sensitivity=uncased, Pre-training strategy=scr2025.084.68
ConfliBERTCase sensitivity=uncased, Pre-training strategy=cont2025.085
BERTCase sensitivity=uncased2025.089.6
BERTCase sensitivity=cased2025.0822.11
PoliBERTweetCase sensitivity=cased, Pre-training strategy=cont2025.08154.42