Research

Papers are roughly grouped by topic below. For a full list, see my Google Scholar or Semantic Scholar.

Reasoning in Multi-Agent Systems

• MAGDi: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models
  Justin Chih-Yao Chen*, Swarnadeep Saha*, Elias Stengel-Eskin, Mohit Bansal
  Pre-print on arXiv, 2024
  [paper] [code] [tweet] [Long]

• ReConcile: Round-Table Conference Improves Reasoning via Consensus among Diverse LLMs
  Justin Chih-Yao Chen, Swarnadeep Saha, and Mohit Bansal
  Pre-print on arXiv, 2023
  [paper] [code] [tweet] [Long]

• Can Language Models Teach Weaker Agents? Teacher Explanations Improve Students via Personalization
  Swarnadeep Saha, Peter Hase, and Mohit Bansal
  Neural Information Processing Systems (NeurIPS), 2023
  [paper] [code] [tweet] [Long] [Poster]

Attributable Generative Reasoning

• Summarization Programs: Interpretable Abstractive Summarization with Neural Modular Trees
  Swarnadeep Saha, Shiyue Zhang, Peter Hase, and Mohit Bansal
  International Conference on Learning Representations (ICLR), 2023
  [paper] [code] [tweet] [Long] [Poster]

• MURMUR: Modular Multi-Step Reasoning for Semi-Structured Data-to-Text Generation
  Swarnadeep Saha, Xinyan Velocity Yu, Mohit Bansal, Ramakanth Pasunuru, and Asli Celikyilmaz
  Annual Meeting of the Association for Computational Linguistics (Findings of ACL), 2023
  [paper] [tweet] [Long] [Poster]

Large Language Model Evaluation

• Branch-Solve-Merge Improves Large Language Model Evaluation and Generation
  Swarnadeep Saha, Omer Levy, Asli Celikyilmaz, Mohit Bansal, Jason Weston, and Xian Li
  Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 2024
  [paper] [tweet] [Long]

• ReCEval: Evaluating Reasoning Chains via Correctness and Informativeness
  Archiki Prasad, Swarnadeep Saha, Xiang Zhou, and Mohit Bansal
  Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023
  [paper] [code] [tweet] [Long] [Poster]

Structured Reasoning over Implicit Knowledge

• Explanation Graph Generation via Pre-trained Language Models: An Empirical Study with Contrastive Learning
  Swarnadeep Saha, Prateek Yadav, and Mohit Bansal
  Annual Meeting of the Association for Computational Linguistics (ACL), 2022
  [paper] [code] [tweet] [Long] [Poster]

• ExplaGraphs: An Explanation Graph Generation Task for Structured Commonsense Reasoning
  Swarnadeep Saha, Prateek Yadav, Lisa Bauer, and Mohit Bansal
  Conference on Empirical Methods in Natural Language Processing (EMNLP), 2021
  [paper] [data/code] [website] [tweet] [Long] [Oral]

Deductive Reasoning

• multiPRover: Generating Multiple Proofs for Improved Interpretability in Rule Reasoning
  Swarnadeep Saha, Prateek Yadav, and Mohit Bansal
  Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 2021
  [paper] [code] [tweet] [Long] [Oral]

• PRover: Proof Generation for Interpretable Reasoning over Rules
  Swarnadeep Saha, Sayan Ghosh, Shashank Srivastava, and Mohit Bansal
  Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020
  [paper] [code] [tweet] [Long] [Oral]

Additional Topics (During PhD)

• Are Hard Examples also Harder to Explain? A Study with Human and Model-Generated Explanations
  Swarnadeep Saha, Peter Hase, Nazneen Rajani, and Mohit Bansal
  Conference on Empirical Methods in Natural Language Processing (EMNLP), 2022
  [paper] [data] [tweet] [Short] [Oral]

• ConjNLI: Natural Language Inference over Conjunctive Sentences
  Swarnadeep Saha, Yixin Nie, and Mohit Bansal
  Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020
  [paper] [data/code] [tweet] [Long] [Poster]

Open Information Extraction

• Open Information Extraction from Conjunctive Sentences
  Swarnadeep Saha, and Mausam
  27th International Conference on Computational Linguistics (COLING), 2018
  [paper] [code] [Long] [Oral]

• Bootstrapping for Numerical Open IE
  Swarnadeep Saha, Harinder Pal, and Mausam
  55th Annual Meeting of the Association for Computational Linguistics (ACL), 2017
  [paper] [code] [Short] [Poster]

Educational NLP

• Pre-Training BERT on Domain Resources for Short Answer Grading
  Chul Sung, Tejas Dhamecha, Swarnadeep Saha, Tengfei Ma, Vinay Reddy, and Rishi Arora
  Conference on Empirical Methods in Natural Language Processing (EMNLP-IJCNLP), 2019
  [paper] [Short] [Poster]

• Aligning Learning Objectives to Learning Resources: A Lexico-Semantic Spatial Approach
  Swarnadeep Saha, Malolan Chetlur, Tejas Indulal Dhamecha, W M Gayathri K Wijayarathna, Red Mendoza, Paul Gagnon, Nabil Zary, and Shantanu Godbole
  28th International Joint Conference on Artificial Intelligence (IJCAI), 2019
  [paper] [Long] [Oral]

• Creating Scoring Rubric from Representative Student Answers for Improved Short Answer Grading
  Smit Marvaniya, Swarnadeep Saha, Tejas I. Dhamecha, Peter Foltz, Renuka Sindhgatta, and Bikram Sengupta
  27th ACM International Conference on Information and Knowledge Management (CIKM), 2018
  [paper] [Long] [Oral]