.png){kind=icon}
.png){kind=icon}
Return
Counterfactual samples constructing and training for commonsense statements estimation
.png){kind=icon}
PRE
AI
C
Z
Z
刘
J
R
D
L
DOI：10.1016/j.ipm.2025.104563.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Introduces CCSG, a novel framework enhancing PE models with counterfactual reasoning. • Leverages SCMs to analyze and mitigate commonsense biases in language model predictions. • Achieves state-of-the-art results on 9 datasets, outperforming GPT-4 and VERA benchmarks.
Journal
I
IF:
6.9
Papers:
5.2K
Citations:
1.4W
Organization
Cited Papers