The reviewed record of science sign in
Pith

arxiv: 2506.09433 · v1 · pith:TLHY7GZL · submitted 2025-06-11 · cs.LG

Mitigating Spurious Correlations in LLMs via Causality-Aware Post-Training

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:TLHY7GZLrecord.jsonopen to challenge →

classification cs.LG
keywords llmscaptcorrelationslanguagespuriousbiasescausality-awareevent
0
0 comments X
read the original abstract

While large language models (LLMs) have demonstrated remarkable capabilities in language modeling, recent studies reveal that they often fail on out-of-distribution (OOD) samples due to spurious correlations acquired during pre-training. Here, we aim to mitigate such spurious correlations through causality-aware post-training (CAPT). By decomposing a biased prediction into two unbiased steps, known as \textit{event estimation} and \textit{event intervention}, we reduce LLMs' pre-training biases without incurring additional fine-tuning biases, thus enhancing the model's generalization ability. Experiments on the formal causal inference benchmark CLadder and the logical reasoning dataset PrOntoQA show that 3B-scale language models fine-tuned with CAPT can outperform both traditional SFT and larger LLMs on in-distribution (ID) and OOD tasks using only 100 ID fine-tuning samples, demonstrating the effectiveness and sample efficiency of CAPT.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.