Boshi Wang (汪博石)

PhD student at The Ohio State University

-------------------------------------

I'm passionate about understanding and creating intelligence systems. Currently I'm focused on understanding & improving the reasoning capabilities of (large) language models.

In my spare time, I like reading & thinking about philosophy, math, and consciousness. Recent favorite book: Gödel, Escher, Bach: an Eternal Golden Braid by Douglas Hofstadter.

About Me

Hi! I'm a 4th-year PhD at The Ohio State University, advised by Prof. Huan Sun. Previously, I received a Bachelor degree in Computer Science at ShanghaiTech University.

News

  • 10/2024 | Our paper studying grokking in Transformer's implicit reasoning is accepted to NeurIPS-24.
  • 05/2023 | Check out our paper on evaluating LLM's reasoning via debate.
  • 10/2022 | Paper on enhancing multi-step reasoning abilities of pretrained language models accepted to EMNLP-22.
  • 08/2021 | Began my PhD @ OSU! Go Buckeyes!

Publications

Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization

B. Wang, X. Yue, Y. Su, H. Sun

NeurIPS-24. Short version in the 1st Mechanistic Interpretability Workshop, ICML-24

Paper Code

LLMs in the Imaginarium: Tool Learning through Simulated Trial and Error

B. Wang, H. Fang, J. Eisner, B. Van Durme, Y. Su

ACL-24, The 62nd Annual Meeting of the Association for Computational Linguistics

Paper Code

Can ChatGPT Defend its Belief in Truth? Evaluting LLM Reasoning via Debate

B. Wang, X. Yue, H. Sun

EMNLP-23 (findings), The 2023 Conference on Empirical Methods in Natural Language Processing

Paper Code

Automatic Evaluation of Attribution by Large Language Models

X. Yue, B. Wang, K. Zhang, Z. Chen, Y. Su, H. Sun

EMNLP-23 (findings), The 2023 Conference on Empirical Methods in Natural Language Processing

Paper Code

Towards Understanding Chain-of-Thought Prompting: An Empirical Study of What Matters

B. Wang, S. Min, X. Deng, J. Shen, W. You, L. Zettlemoyer, H. Sun

ACL-23 (Honorable Mention), The 61st Annual Meeting of the Association for Computational Linguistics

Paper Code

A Retrieve-and-Read Framework for Knowledge Graph Link Prediction

V. Pahuja, B. Wang, H. Latapie, J. Srinivasa, Y. Su

CIKM-23, The 32nd ACM International Conference on Information and Knowledge Management

Paper

Mind2Web: Towards a Generalist Agent for the Web

X. Deng, Y. Gu, B. Zheng, S. Chen, S. Stevens, B. Wang, H. Sun, Y. Su

NeurIPS-23 Dataset Track (Spotlight), Thirty-seventh Conference on Neural Information Processing Systems

Paper Code

Iteratively Prompt Pre-trained Language Models for Chain of Thought

B. Wang, X. Deng, H. Sun

EMNLP-22 (Oral), The 2022 Conference on Empirical Methods in Natural Language Processing

Paper Code

Automatic Loss Function Search for Predict-Then-Optimize Problems with Strong Ranking Property

B. Wang*, J. Yi*, H. Dong, B. Qiao, C. Luo, Q. Lin

ICLR-22, The Tenth International Conference on Learning Representations

Paper

Homomorphic Sensing: Sparsity and Noise

L. Peng, B. Wang and M. C. Tsakiris

ICML-21, Proceedings of the 38th International Conference on Machine Learning

Paper

Designing Context-Sensitive Norm Inverse Reinforcement Learning Framework for Norm-Compliant Autonomous Agents

Y. Guo, B. Wang, D. Hughes, M. Lewis, K. Sycara

ROMAN-20, 29th IEEE International Conference on Robot and Human Interactive Communication

Paper

Education

  • 08/2021 - Present | PhD, Ohio State University, USA
  • 06/2019 - 09/2019 | Robotics Institute Summer Scholars, Carnegie Mellon University, USA
  • 09/2016 - 07/2020 | B.S.E in Computer Science, ShanghaiTech University, CHN

Work/Intern Experience

06/2023 - 08/2023

Microsoft Semantic Machines

Research Intern

01/2021 - 06/2021

Microsoft Research Asia

Research Intern

07/2020 - 12/2020

Institue of Mathematical Sciences, ShanghaiTech University

Research Assistant

Service

  • Program Committee/Reviewer: NeurIPS (Top Reviewers 2024), ICLR, ACL-ARR, AAAI