Bhavana Dalvi Mishra

Lead Research Scientist
Allen Institute for AI
Seattle, WA, USA

Email: bhavanad AT allenai DOT org


About Me

I am a Lead Research Scientist at AI2. My research interests are in the area of Generative AI, Interactive reasoning and Scientific Discovery. I received my Ph.D. in Computer Science from the Carnegie Mellon University in 2015, advised by Prof. William Cohen and Prof. Jamie Callan. I completed Masters in Computer Science from the Indian Institute of Technology, Bombay in 2007. Prior to PhD, I worked at Google R&D Bangalore for two years. Besides research, I love hiking, meditation, and music.

Selected Publications

The complete list of publications can be seen from my Google Scholar page.
* indicates equal contribution

Generative AI agents for Discovery

  • DiscoveryBench: Towards Data-Driven Discovery with Large Language Models [Paper] [Website]
    Bodhisattwa Prasad Majumder*, Harshit Surana*, Dhruv Agarwal*, Bhavana Dalvi Mishra*, Abhijeetsingh Meena, Aryan Prakhar, Tirth Vora, Tushar Khot, Ashish Sabharwal, Peter Clark
    arXiv.
  • DISCOVERYWORLD: A Virtual Environment for Developing and Evaluating Automated Scientific Discovery Agents [Paper] [Website]
    Peter Jansen, Marc-Alexandre Côté, Tushar Khot, Erin Bransom, Bhavana Dalvi Mishra, Bodhisattwa Prasad Majumder, Oyvind Tafjord, Peter Clark
    arXiv.

Continual Learning with Interaction

  • Skill Set Optimization: Reinforcing Language Model Behavior via Transferable Skills [Paper] [Website]
    Kolby Nottingham, Bodhisattwa Prasad Majumder*, Bhavana Dalvi Mishra*, Sameer Singh, Peter Clark, Roy Fox
    ICML 2024.
  • CLIN: A Continually Learning Language Agent for Rapid Task Adaptation and Generalization [Paper] [Website]
    Bodhisattwa Prasad Majumder, Bhavana Dalvi Mishra, Peter Jansen, Oyvind Tafjord, Niket Tandon, Li Zhang, Chris Callison-Burch, Peter Clark
    COLM 2024.

Explainable Teachable Reasoning

  • Towards Teachable Reasoning Systems: Using a Dynamic Memory of User Feedback for Continual System Improvement [Paper] [Dataset] [Blog]
    Bhavana Dalvi Mishra, Oyvind Tafjord, Peter Clark
    EMNLP 2022.
  • Entailer: Answering Questions with Faithful and Truthful Chains of Reasoning
    [Paper] [Data] [Code] [Blog]
    Oyvind Tafjord, Bhavana Dalvi Mishra, Peter Clark
    EMNLP 2022.
  • Explaining Answers with Entailment Trees [Paper] [Data] [Code]
    Bhavana Dalvi Mishra*, Peter Jansen*, Oyvind Tafjord, Zhengnan Xie, Hannah Smith, Leighanna Pipatanangkura, Peter Clark
    EMNLP 2021.
  • ProofWriter: Generating Implications, Proofs, and Abductive Statements over Natural Language [Paper] [Data] [Demo]
    Oyvind Tafjord, Bhavana Dalvi Mishra, Peter Clark
    Findings of ACL 2021.

Mental Models behind Language Models

  • Do language models have coherent mental models of everyday things? [Paper] [Website] [Blog]
    Yuling Gu, Bhavana Dalvi Mishra, Peter Clark
    ACL 2023.
  • Just-DREAM-about-it: Figurative Language Understanding with DREAM-FLUTE [Paper] [Blog]
    Yuling Gu, Yao Fu, Valentina Pyatkin, Ian Magnusson, Bhavana Dalvi Mishra, Peter Clark
    The Third Workshop on Figurative Language Processing at EMNLP 2022.
  • DREAM: Improving Situational QA by First Elaborating the Situation [Paper] [Website] [Blog]
    Yuling Gu, Bhavana Dalvi Mishra, Peter Clark
    NAACL 2022.

Procedural Reasoning

  • A Dataset for Tracking Entities in Open Domain Procedural Text [Paper] [Data] [Code]
    Niket Tandon, Keisuke Sakaguchi, Bhavana Dalvi Mishra, Dheeraj Rajagopal, Peter Clark, Michal Guerquin, Kyle Richardson, Eduard Hovy
    EMNLP 2020.
  • Everything Happens for a Reason: Discovering the Purpose of Actions in Procedural Text [Paper]
    Bhavania Dalvi Mishra*, Niket Tandon*, Antoine Bosselut, Wen-tau Yih, Peter Clark
    EMNLP 2020.
  • WIQA: A dataset for "What if..." reasoning over procedural text [Paper] [Data] [Code]
    Niket Tandon*, Bhavana Dalvi Mishra*, Keisuke Sakaguchi, Antoine Bosselut, Peter Clark
    EMNLP 2019.
  • Reasoning about Actions and State Changes by Injecting Commonsense Knowledge [Paper] [Data] [Code]
    Niket Tandon*, Bhavana Dalvi Mishra*, Joel Grus, Wen-tau Yih, Antoine Bosselut, Peter Clark
    EMNLP 2018.