Mitigating Spurious Correlations in NLI via LLM-Synthesized Counterfactuals and Dynamic Balanced Sampling
Natural Language Inference (NLI) models frequently rely on spurious correlations rather than semantic reasoning. Existing mitigation strategies often incur high annotation costs or trigger catastrophic forgetting during fine-tuning. We propose an automated, scalable pipeline to address these limitations. First, we introduce Log-Frequency LMI (LF-LMI) to accurately detect semantic artifacts. Second, we generate a high-quality synthetic contrast set via an LLM-synthesis pipeline with multi-judge verification. Finally, we introduce Dynamic Balanced Sampling, a training strategy that rotates the original data distribution to prevent forgetting. Our method improves consistency on a challenging benchmark from 63.5% to 81.0% while maintaining 88.4% in-domain accuracy, significantly outperforming naive fine-tuning.
Code (0)
등록된 구현이 없습니다.
Tasks
Natural Language InferenceSimilar Papers 제목 키워드 기반
Uncovering Main Causalities for Long-tailed Information Extraction
Information Extraction (IE) aims to extract structural information from unstructured texts. In practice, long-tailed distributions caused by the selection bias of a dataset, may lead to incorrect correlations, also known…
Causal InferencecounterfactualSelection biasMitigating Clever Hans Strategies in Image Classifiers through Generating Counterexamples
Deep learning models remain vulnerable to spurious correlations, leading to so-called Clever Hans predictors that undermine robustness even in large-scale foundation and self-supervised models. Group distributional robus…
Knowledge DistillationMitigating Shortcut Learning with Diffusion Counterfactuals and Diverse Ensembles
Spurious correlations in the data, where multiple cues are predictive of the target labels, often lead to a phenomenon known as shortcut learning, where a model relies on erroneous, easy-to-learn cues while ignoring reli…
DiversityRaVL: Discovering and Mitigating Spurious Correlations in Fine-Tuned Vision-Language Models
Fine-tuned vision-language models (VLMs) often capture spurious correlations between image features and textual attributes, resulting in degraded zero-shot performance at test time. Existing approaches for addressing spu…
image-classificationImage Classificationzero-shot-classificationZero-Shot LearningVisual Disentangled Diffusion Autoencoders: Scalable Counterfactual Generation for Foundation Models
Foundation models, despite their robust zero-shot capabilities, remain vulnerable to spurious correlations and 'Clever Hans' strategies. Existing mitigation methods often rely on unavailable group labels or computational…
Knowledge Distillation