no code implementations • 1 Apr 2024 • Pengda Wang, Zilin Xiao, Hanjie Chen, Frederick L. Oswald
It can occur even when LLMs possess the correct knowledge while failing in a cognitive trap.
1 code implementation • 28 Feb 2024 • Hanjie Chen, Zhouxiang Fang, Yash Singla, Mark Dredze
To address these challenges, we construct two new datasets: JAMA Clinical Challenge and Medbullets.
no code implementations • 28 Feb 2024 • Zhengping Jiang, Yining Lu, Hanjie Chen, Daniel Khashabi, Benjamin Van Durme, Anqi Liu
This is achieved by assessing the conditional V-information \citep{hewitt-etal-2021-conditional} with a predictive family robust against leaky features that can be exploited by a small model.
no code implementations • 2 Sep 2023 • Haiyan Zhao, Hanjie Chen, Fan Yang, Ninghao Liu, Huiqi Deng, Hengyi Cai, Shuaiqiang Wang, Dawei Yin, Mengnan Du
For each paradigm, we summarize the goals and dominant approaches for generating local explanations of individual predictions and global explanations of overall model knowledge.
1 code implementation • 3 Feb 2023 • Arshdeep Sekhon, Hanjie Chen, Aman Shrivastava, Zhe Wang, Yangfeng Ji, Yanjun Qi
Recent NLP literature has seen growing interest in improving model interpretability.
no code implementations • 19 Dec 2022 • Aaron Chan, Zhiyuan Zeng, Wyatt Lake, Brihi Joshi, Hanjie Chen, Xiang Ren
First, KNIFE finetunes a teacher LM (given task input and FTR) to predict the task output, transferring reasoning knowledge from the FTRs to the teacher's hidden states.
no code implementations • 10 Dec 2022 • Ruixuan Tang, Hanjie Chen, Yangfeng Ji
Some recent works observed the instability of post-hoc explanations when input side perturbations are applied to the model.
1 code implementation • 10 Oct 2022 • Hanjie Chen, Faeze Brahman, Xiang Ren, Yangfeng Ji, Yejin Choi, Swabha Swayamdipta
More concretely, we propose a metric called REV (Rationale Evaluation with conditional V-information), to quantify the amount of new, label-relevant information in a rationale beyond the information already available in the input or the label.
1 code implementation • 19 May 2022 • Wanyu Du, Hanjie Chen, Yangfeng Ji
In task-oriented dialogue systems, response generation from meaning representations (MRs) often suffers from limited training examples, due to the high cost of annotating MR-to-Text pairs.
no code implementations • insights (ACL) 2022 • Hanjie Chen, Guoqing Zheng, Ahmed Hassan Awadallah, Yangfeng Ji
Although adapting pre-trained language models with few examples has shown promising performance on text classification, there is a lack of understanding of where the performance gain comes from.
1 code implementation • 23 Mar 2022 • Hanjie Chen, Yangfeng Ji
Neural language models show vulnerability to adversarial examples which are semantically similar to their original counterparts with a few words replaced by their synonyms.
no code implementations • 11 Jan 2022 • Hanjie Chen, Wanyu Du, Yangfeng Ji
Explaining predictive uncertainty is an important complement to explaining prediction labels in helping users understand model decision making and gaining their trust on model predictions, while has been largely ignored in prior works.
1 code implementation • EMNLP (BlackboxNLP) 2021 • Sanchit Sinha, Hanjie Chen, Arshdeep Sekhon, Yangfeng Ji, Yanjun Qi
Via a small portion of word-level swaps, these adversarial perturbations aim to make the resulting text semantically and spatially similar to its seed input (therefore sharing similar interpretations).
1 code implementation • NAACL 2021 • Hanjie Chen, Song Feng, Jatin Ganhotra, Hui Wan, Chulaka Gunasekara, Sachindra Joshi, Yangfeng Ji
Most existing methods generate post-hoc explanations for neural network models by identifying individual feature attributions or detecting interactions between adjacent features.
3 code implementations • EMNLP 2020 • Hanjie Chen, Yangfeng Ji
To build an interpretable neural text classifier, most of the prior work has focused on designing inherently interpretable models or finding faithful explanations.
2 code implementations • ACL 2020 • Hanjie Chen, Guangtao Zheng, Yangfeng Ji
Experiments show the effectiveness of the proposed method in providing explanations that are both faithful to models and interpretable to humans.
no code implementations • 10 Sep 2019 • Hanjie Chen, Yangfeng Ji
Experiments show the proposed data augmentation methods significantly improve the explainability of both neural classifiers.