Publications | Natural Language Processing

Preprints

• Did I Faithfully Say What I Thought? Bridging the Gap Between Neural Activity and Self-Explanations in Large Language Models
  Milan Bhan, Jean-Noel Vittaut, Nicolas Chesneau, Sarath Chandar, and Marie-Jeanne Lesot
  In ArXiv, 2025.
  #NLP
  [arXiv]

• Boosting LLM Reasoning via Spontaneous Self-Correction
  Xutong Zhao, Tengyu Xu, Xuewei Wang, Zhengxing Chen, Di Jin, Liang Tan, Yen-Ting, Zishun Yu, Zhuokai Zhao, Yun He, Sinong Wang, Han Fang, Sarath Chandar, and Chen Zhu
  In ArXiv, 2025.
  #NLP, #RL
  [arXiv]

• Steering Large Language Model Activations in Sparse Spaces
  Reza Bayat, Ali Rahimi-Kalahroudi, Mohammad Pezeshki, Sarath Chandar, and Pascal Vincent
  In ArXiv, 2025.
  #NLP, #DL
  [arXiv]

• Too Big to Fool: Resisting Deception in Language Models
  Mohammad Reza Samsami, Mats Leon Richter, Juan Rodriguez, Megh Thakkar, Sarath Chandar, and Maxime Gasse
  In ArXiv, 2024.
  #NLP
  [arXiv]

• Interpretability Needs a New Paradigm
  Andreas Madsen, Himabindu Lakkaraju, Siva Reddy, and Sarath Chandar
  In ArXiv, 2024.
  #NLP, #DL
  [arXiv]

Conference and Journal Papers

2025

32. Combining Domain and Alignment Vectors Provides Better Knowledge-Safety Trade-offs in LLMs
    Megh Thakkar, Quentin Fournier, Matthew Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, and Sarath Chandar
    Annual Meeting of the Association for Computational Linguistics (ACL), 2025.
    #NLP
    [arXiv]

33. Small Encoders Can Rival Large Decoders in Detecting Groundedness
    Istabrak Abbes, Gabriele Prato, Quentin Fournier, Fernando Rodriguez, Alaa Boukhary, Adam Elwood, and Sarath Chandar
    Findings of the Association for Computational Linguistics (ACL), 2025.
    #NLP
    

34. Do Robot Snakes Dream like Electric Sheep? Investigating the Effects of Architectural Inductive Biases on Hallucination
    Jerry Huang, Prasanna Parthasarathi, Mehdi Rezagholizadeh, Boxing Chen, and Sarath Chandar
    Findings of the Association for Computational Linguistics (ACL), 2025.
    #NLP
    [arXiv]

35. IDAT: A Multi-Modal Dataset and Toolkit for Building and Evaluating Interactive Task-Solving Agents
    Shrestha Mohanty, Negar Arabzadeh, Andrea Tupini, Yuxuan Sun, Alexey Skrynnik, Artem Zholus, Marc-Alexandre Côté, and Julia Kiseleva
    ACM SIGIR Conference on Research and Development in Information Retrieval, 2025.
    #NLP
    [arXiv]

36. NeoBERT: A Next Generation BERT
    Lola Le Breton, Quentin Fournier, Mariam El Mezouar, and Sarath Chandar
    Transactions on Machine Learning Research (TMLR), 2025.
    #NLP
    [openreview]

37. ChartGemma: Visual Instruction-tuning for Chart Reasoning in the Wild
    Ahmed Masry*, Megh Thakkar*, Aayush Bajaj, Aaryaman Kartha, Enamul Hoque, and Shafiq Joty
    International Conference on Computational Linguistics (COLING) Industry Track, 2025.
    #NLP
    [acl], [arXiv], [code]

2024

26. WorkArena++: Towards Compositional Planning and Reasoning-based Common Knowledge Work Tasks
    Leo Boisvert*, Megh Thakkar*, Maxime Gasse, Massimo Caccia, Thibault Le Sellier De Chezelles, Quentin Cappart, Nicolas Chapados, Alexandre Lacoste, and Alexandre Drouin
    Conference on Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track, 2024.
    #NLP
    [openreview], [arXiv], [code]

27. Exploring Quantization for Efficient Pre-Training of Transformer Language Models
    Kamran Chitsaz, Quentin Fournier, Gonçalo Mordido, and Sarath Chandar
    Findings of the Association for Computational Linguistics (EMNLP), 2024.
    #NLP, #DL
    [acl], [arXiv]

28. Context-Aware Assistant Selection for Improved Inference Acceleration with Large Language Models
    Jerry Huang, Prasanna Parthasarathi, Mehdi Rezagholizadeh, and Sarath Chandar
    Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024.
    #NLP
    [acl], [arXiv]

29. Do Large Language Models Know How Much They Know?
    Gabriele Prato, Jerry Huang, Prasanna Parthasarathi, Shagun Sodhani, and Sarath Chandar
    Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024.
    #NLP
    [acl]

30. Should We Attend More or Less? Modulating Attention for Fairness
    Abdelrahman Zayed, Gonçalo Mordido, Samira Shabanian, and Sarath Chandar
    Conference on Language Modeling (COLM), 2024.
    #NLP
    [openreview], [arXiv]

31. Are self-explanations from Large Language Models faithful?
    Andreas Madsen, Sarath Chandar, and Siva Reddy
    Findings of the Association for Computational Linguistics (ACL), 2024.
    #NLP
    [acl], [arXiv], [code], [YouTube]

32. A deep-dive into the tradeoffs of preference alignment with PEFT
    Megh Thakkar, Quentin Fournier, Matthew Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, and Sarath Chandar
    Annual Meeting of the Association for Computational Linguistics (ACL), 2024.
    #NLP
    [acl], [arXiv]

33. Why Don’t Prompt-Based Fairness Metrics Correlate?
    Abdelrahman Zayed, Gonçalo Mordido, Ioana Baldini, and Sarath Chandar
    Annual Meeting of the Association for Computational Linguistics (ACL), 2024.
    #NLP
    [acl], [arXiv], [YouTube]

34. Sub-goal Distillation: A Method to Improve Small Language Agents
    Maryam Hashemzadeh, Elias Stengel-Eskin, Sarath Chandar, and Marc-Alexandre Cote
    Conference on Lifelong Learning Agents (CoLLAs), 2024. [Oral presentation.]
    #RL, #NLP
    [arXiv]

35. Faithfulness Measurable Masked Language Models
    Andreas Madsen, Siva Reddy, and Sarath Chandar
    International Conference on Machine Learning (ICML), 2024. [Spotlight award - top 3.5%]
    #NLP
    [pmlr], [arXiv], [code], [YouTube], [blogpost]

36. MVP: Minimal Viable Phrase for Long Text Understanding
    Louis Clouâtre, Amal Zouaq, and Sarath Chandar
    Joint International Conference on Computational Linguistics, Language, Resources and Evaluation (LREC-COLING), 2024.
    #NLP
    [acl]

37. Fairness-Aware Structured Pruning in Transformers
    Abdelrahman Zayed, Gonçalo Mordido, Samira Shabanian, Ioana Baldini, and Sarath Chandar
    AAAI Conference on Artificial Intelligence (AAAI), 2024.
    #NLP
    [aaai], [arXiv], [YouTube]

2023

14. Self-Influence Guided Data Reweighting for Language Model Pre-training
    Megh Thakkar, Tolga Bolukbasi, Sriram Ganapathy, Shikhar Vashishth, Sarath Chandar, and Partha Talukdar
    Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.
    #NLP
    [acl], [openreview], [arXiv]

15. EpiK-Eval: Evaluation for Language Models as Epistemic Models
    Gabriele Prato, Jerry Huang, Prasanna Parthasarathi, Shagun Sodhani, and Sarath Chandar
    Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.
    #NLP
    [acl], [openreview], [arXiv], [code]

16. Measuring the Knowledge Acquisition-Utilization Gap in Pretrained Language Models
    Amirhossein Kazemnejad, Mehdi Rezagholizadeh, Prasanna Parthasarathi, and Sarath Chandar
    Findings of the Association for Computational Linguistics (EMNLP), 2023.
    #NLP
    [acl], [openreview], [arXiv]

17. Deep Learning on a Healthy Data Diet: Finding Important Examples for Fairness
    Abdelrahman Zayed, Prasanna Parthasarathi, Gonçalo Mordido, Hamid Palangi, Samira Shabanian, and Sarath Chandar
    AAAI Conference on Artificial Intelligence (AAAI), 2023.
    #NLP
    [aaai], [arXiv], [YouTube]

2022

10. Evaluating the Faithfulness of Importance Measures in NLP by Recursively Masking Allegedly Important Tokens and Retraining
    Andreas Madsen, Nicholas Meade, Vaibhav Adlakha, and Siva Reddy
    Findings of the Association for Computational Linguistics (EMNLP), 2022.
    [BlackboxNLP Workshop, 2022]
    #NLP
    [acl], [arXiv], [code]

11. Detecting Languages Unintelligible to Multilingual Models through Local Structure Probes
    Louis Clouâtre, Prasanna Parthasarathi, Amal Zouaq, and Sarath Chandar
    Findings of the Association for Computational Linguistics (EMNLP), 2022.
    #NLP
    [acl]

12. Local Structure Matters Most in Most Languages
    Louis Clouâtre, Prasanna Parthasarathi, Amal Zouaq, and Sarath Chandar
    Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the International Joint Conference on Natural Language Processing (AACL-IJCNLP), 2022.
    #NLP
    [acl]

13. Post-hoc Interpretability for Neural NLP: A Survey
    Andreas Madsen, Siva Reddy, and Sarath Chandar
    ACM Computing Surveys, 2022.
    #NLP
    [acm], [arXiv]

14. Local Structure Matters Most: Perturbation Study in NLU
    Louis Clouâtre, Prasanna Parthasarathi, Amal Zouaq, and Sarath Chandar
    Findings of the Association for Computational Linguistics (ACL), 2022.
    #NLP
    [acl], [arXiv]

2021

5. Benchmarking Bias Mitigation Algorithms in Representation Learning through Fairness Metrics
   Charan Reddy, Deepak Sharma, Soroush Mehri, Adriana Romero, Samira Shabanian, and Sina Honari
   Conference on Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track, 2021.
   #NLP
   [neurips], [openreview], [code]

6. A Survey of Data Augmentation Approaches for NLP
   Steven Y. Feng, Varun Gangal, Jason Wei, Sarath Chandar, Soroush Vosoughi, Teruko Mitamura, and Eduard Hovy
   Findings of the Association for Computational Linguistics (ACL-IJCNLP), 2021.
   #NLP
   [acl], [arXiv]

7. MLMLM: Link Prediction with Mean Likelihood Masked Language Model
   Louis Clouâtre, Philippe Trempe, Amal Zouaq, and Sarath Chandar
   Findings of the Association for Computational Linguistics (ACL-IJCNLP), 2021.
   #NLP
   [acl], [arXiv]

8. A Brief Study on the Effects of Training Generative Dialogue Models with a Semantic loss
   Prasanna Parthasarathi, Mohamed Abdelsalam, Joelle Pineau, and Sarath Chandar
   Proceedings of the 22nd Annual SIGdial Meeting on Discourse and Dialogue, 2021.
   #NLP
   [acl]

9. Do Encoder Representations of Generative Dialogue Models Encode Sufficient Information about the Task ?
   Prasanna Parthasarathi, Sarath Chandar, and Joelle Pineau
   Proceedings of the 22nd Annual SIGdial Meeting on Discourse and Dialogue, 2021.
   #NLP
   [acl]