CREST: A causal framework for mitigating shortcut learning in language models through counterfactual reasoning.

Zhonghua Liu|Wei Shao|Shaolong E.|Xia Cao


Anthology ID:DBLP:journals/ipm/LiuSEC26
Year:2026
Venue:Information Processing and Management
Pages:104418
URL:https://doi.org/10.1016/J.IPM.2025.104418
DOI:https://doi.org/10.1016/J.IPM.2025.104418
DBLP:journals/ipm/LiuSEC26
BibTeX:
@article{liu-2026-crest, author = {Zhonghua Liu and Wei Shao and Shaolong E. and Xia Cao}, title = {{CREST: A causal framework for mitigating shortcut learning in language models through counterfactual reasoning}}, journal = {Information Processing and Management}, volume = {63}, number = {2}, pages = {104418}, year = {2026}, url = {https://doi.org/10.1016/J.IPM.2025.104418}, doi = {https://doi.org/10.1016/J.IPM.2025.104418} }