Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Language Modeling on Model-generated outputs

5.45PPL (Mistral-7B)

Mean-AcT

5.37045.90776.4456.9823Oct 5, 2025
Updated 20d ago

Evaluation Results

MethodLinks
2025.10
5.45
2025.10
5.46
2025.10
5.5
2025.10
5.56
2025.10
5.68
2025.10
5.71
2025.10
5.85
2025.10
6.05
2025.10
6.08
2025.10
6.17
2025.10
6.67
2025.10
6.68
2025.10
6.81
2025.10
6.85
2025.10
6.97
2025.10
7.02
2025.10
7.04
2025.10
7.13
2025.10
7.24
2025.10
7.44