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

Reliability of post-edit LLMs on Wikipedia

100BLEU

GRACE

71.9279.2186.593.79Jun 9, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
100100
2025.06
97.597.7
2025.06
96.696.5
2025.06
96.296.6
2025.06
95.495.9
2025.06
94.695.7
2025.06
94.194.6
2025.06
92.593.7
2025.06
92.293.1
2025.06
9293
2025.06
91.492.5
2025.06
91.293.1
2025.06
90.792.2
2025.06
90.691.6
2025.06
90.491.6
2025.06
90.491.7
2025.06
88.789.9
2025.06
87.889.5
2025.06
87.589.2
2025.06
86.988.3
2025.06
86.688.2
2025.06
83.786.2
2025.06
83.585
2025.06
8384.7
2025.06
8283.5
2025.06
74.976.3
2025.06
73.475.7
2025.06
7374.7