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

Factuality Evaluation on Held-out Dataset

8.84Factuality Score

F-DPO

4.85685.89096.9257.9591Jan 6, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.01
8.840.8
2026.01
8.266.8
2026.01
8.049.2
2026.01
87.2
2026.01
7.98
2026.01
7.98.4
2026.01
7.87.2
2026.01
7.68.2
2026.01
7.59.8
2026.01
7.311.6
2026.01
7.114.2
2026.01
715.4
2026.01
6.9518.2
2026.01
6.523.8
2026.01
6.425.8
2026.01
6.1430.2
2026.01
629
2026.01
5.830
2026.01
5.2642.4
2026.01
5.0240
2026.01
5.0143.2