AGR: Reinforced causal agent-guided self-explaining rationalization
Y Zhao, Z Wang, X Li, J Liang, R Li - Proceedings of the 62nd …, 2024 - aclanthology.org
Most existing rationalization approaches are susceptible to degeneration accumulation due
to a lack of effective control over the learning direction of the model during training. To …
to a lack of effective control over the learning direction of the model during training. To …
Mare: Multi-aspect rationale extractor on unsupervised rationale extraction
Unsupervised rationale extraction aims to extract text snippets to support model predictions
without explicit rationale annotation. Researchers have made many efforts to solve this task …
without explicit rationale annotation. Researchers have made many efforts to solve this task …
Is the MMI Criterion Necessary for Interpretability? Degenerating Non-causal Features to Plain Noise for Self-Rationalization
An important line of research in the field of explainability is to extract a small subset of crucial
rationales from the full input. The most widely used criterion for rationale extraction is the …
rationales from the full input. The most widely used criterion for rationale extraction is the …
A Unified Causal View of Instruction Tuning
Instruction tuning on a mixture of tasks has improved zero-shot capabilities in natural
language processing (NLP). Nevertheless, existing methods often learn features that exhibit …
language processing (NLP). Nevertheless, existing methods often learn features that exhibit …
Interlocking-free Selective Rationalization Through Genetic-based Learning
F Ruggeri, G Signorelli - arXiv preprint arXiv:2412.10312, 2024 - arxiv.org
A popular end-to-end architecture for selective rationalization is the select-then-predict
pipeline, comprising a generator to extract highlights fed to a predictor. Such a cooperative …
pipeline, comprising a generator to extract highlights fed to a predictor. Such a cooperative …
Adversarial Attack for Explanation Robustness of Rationalization Models
Rationalization models, which select a subset of input text as rationale—crucial for humans
to understand and trust predictions—have recently emerged as a prominent research area …
to understand and trust predictions—have recently emerged as a prominent research area …