Sample-Efficient Online Learning in LM Agents via Hindsight Trajectory Rewriting
Michael Y. Hu, Ben Van Durme, Jacob Andreas, Harsh Jhamtani
October 2025
Copilot Tuning Research and Applied Science
论文与出版物
-
-
-
LM Agents for Coordinating Multi-User Information Gathering
Harsh Jhamtani, Jacob Andreas, Ben Van Durme
ACL | July 2025
-
Direct Reasoning Optimization: LLMs Can Reward And Refine Their Own Reasoning for Open-Ended Tasks
Yifei Xu, Tusher Chakraborty, Srinagesh Sharma, Leonardo Nunes, Emre Kiciman, Songwu Lu, Ranveer Chandra
Arxiv | June 2025
-
Walk the Talk? Measuring the Faithfulness of Large Language Model Explanations
Katie Matton, Robert Osazuwa Ness, John Guttag, Emre Kiciman
2025 International Conference on Learning Representations | April 2025
-
Steering Large Language Models between Code Execution and Textual Reasoning
Yongchao Chen, Harsh Jhamtani, Srinagesh Sharma, Chuchu Fan, Chi Wang
ICLR | April 2025
-
RLTHF: Targeted Human Feedback for LLM Alignment
Yifei Xu, Tusher Chakraborty, Emre Kiciman, Bibek Aryal, Eduardo Rodrigues, Srinagesh Sharma, Roberto Estevao, Maria Angels de Luis Balaguer, Jessica Wolk, Rafael Padilha, Leonardo Nunes, Shobana Balakrishna, Songwu Lu, Ranveer Chandra
ICML’25 | February 2025
-
-
-
Language-to-Code Translation with a Single Labeled Example
Kaj Bostrom, Harsh Jhamtani, Hao Fang, Sam Thomson, Richard Shin, Patrick Xia, Ben Van Durme, Jason Eisner, Jacob Andreas
EMNLP | November 2024
-
Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots
Margaret Capetz, Swati Sharma, Rafael Padilha, Peder Olsen, Jessica Wolk, Emre Kiciman, Ranveer Chandra
ArXiv | November 2024, 第 abs/2411.16872 卷
-
Artificial Intelligence–Based Copilots to Generate Causal Evidence
Maya Petersen, Ahmed Alaa, Emre Kiciman, Chris Holmes, Mark van der Laan
NEJM AI | November 2024
-
Ontologically Faithful Generation of Non-Player Character Dialogues
Nathaniel Weir, Ryan Thomas, Randolph D'Amore, Kellie Hill, Ben Van Durme, Harsh Jhamtani
EMNLP | November 2024
-
Learning to Retrieve Iteratively for In-Context Learning
Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Ben Van Durme
EMNLP | November 2024
Outstanding Paper Award
-
Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Pass
Tong Chen, Hao Fang, Patrick Xia, Xiaodong Liu, Ben Van Durme, Luke Zettlemoyer, Jianfeng Gao, Hao Cheng
ICLR 2025 | November 2024
-
Multi-Field Adaptive Retrieval
Millicent Li, Tongfei Chen, Ben Van Durme, Patrick Xia
ICLR 2025 | October 2024
Spotlight
-
Causal Reasoning and Large Language Models: Opening a New Frontier for Causality
Emre Kiciman, Robert Osazuwa Ness, Amit Sharma, Chenhao Tan
Transactions on Machine Learning Research (TMLR) | August 2024
Outstanding Certification Finalist
Selected for presentation at ICLR 2025
-
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani, Hao Fang, Patrick Xia, Eran Levy, Jacob Andreas, Ben Van Durme
IJCAI 2024 | August 2024
-
Large Language Models as Co-Pilots for Causal Inference in Medical Studies
Ahmed Alaa, Rachael V. Phillips, Emre Kiciman, Laura B. Balzer, M. V. D. Laan, Maya Petersen
ArXiv | July 2024, 第 abs/2407.19118 卷
-
Interpreting User Requests in the Context of Natural Language Standing Instructions
Nikita Moghe, Patrick Xia, Jacob Andreas, Jason Eisner, Ben Van Durme, Harsh Jhamtani
NAACL | June 2024
-
Defending Against Indirect Prompt Injection Attacks With Spotlighting
Keegan Hines, Gary Lopez, Matthew Hall, Federico Zarfati, Yonatan Zunger, Emre Kiciman
ArXiv | March 2024, 第 abs/2403.14720 卷
-
-
-
Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models
Jingwei Yi, Yueqi Xie, Bin Benjamin Zhu, Emre Kiciman, Guangzhong Sun, Xing Xie, Fangzhao Wu
ArXiv | December 2023, 第 abs/2312.14197 卷
-
GPT-4 as an Agronomist Assistant? Answering Agriculture Exams Using Large Language Models
Bruno Silva, Leonardo Nunes, Roberto Estevão, Vijay Aski, Ranveer Chandra
October 2023
-
SCREWS: A Modular Framework for Reasoning with Revisions
Kumar Shridhar, Harsh Jhamtani, Hao Fang, Ben Van Durme, Jason Eisner, Patrick Xia
arXiv: Computation and Language | September 2023
-
The Whole Truth and Nothing But the Truth: Faithful and Controllable Dialogue Response Generation with Dataflow Transduction and Constrained Decoding
Hao Fang, Anusha Balakrishnan, Harsh Jhamtani, John Bufe, Jean Crawford, Jayant Krishnamurthy, Adam Pauls, Jason Eisner, Jacob Andreas, Dan Klein
Findings of ACL 2023 | July 2023
-
-
-
Sample-Efficient Online Learning in LM Agents via Hindsight Trajectory Rewriting
Michael Y. Hu, Ben Van Durme, Jacob Andreas, Harsh Jhamtani
October 2025
-
LM Agents for Coordinating Multi-User Information Gathering
Harsh Jhamtani, Jacob Andreas, Ben Van Durme
ACL | July 2025
-
Direct Reasoning Optimization: LLMs Can Reward And Refine Their Own Reasoning for Open-Ended Tasks
Yifei Xu, Tusher Chakraborty, Srinagesh Sharma, Leonardo Nunes, Emre Kiciman, Songwu Lu, Ranveer Chandra
Arxiv | June 2025
-
Walk the Talk? Measuring the Faithfulness of Large Language Model Explanations
Katie Matton, Robert Osazuwa Ness, John Guttag, Emre Kiciman
2025 International Conference on Learning Representations | April 2025
-
Steering Large Language Models between Code Execution and Textual Reasoning
Yongchao Chen, Harsh Jhamtani, Srinagesh Sharma, Chuchu Fan, Chi Wang
ICLR | April 2025
-
RLTHF: Targeted Human Feedback for LLM Alignment
Yifei Xu, Tusher Chakraborty, Emre Kiciman, Bibek Aryal, Eduardo Rodrigues, Srinagesh Sharma, Roberto Estevao, Maria Angels de Luis Balaguer, Jessica Wolk, Rafael Padilha, Leonardo Nunes, Shobana Balakrishna, Songwu Lu, Ranveer Chandra
ICML’25 | February 2025
-
Language-to-Code Translation with a Single Labeled Example
Kaj Bostrom, Harsh Jhamtani, Hao Fang, Sam Thomson, Richard Shin, Patrick Xia, Ben Van Durme, Jason Eisner, Jacob Andreas
EMNLP | November 2024
-
Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots
Margaret Capetz, Swati Sharma, Rafael Padilha, Peder Olsen, Jessica Wolk, Emre Kiciman, Ranveer Chandra
ArXiv | November 2024, 第 abs/2411.16872 卷
-
Artificial Intelligence–Based Copilots to Generate Causal Evidence
Maya Petersen, Ahmed Alaa, Emre Kiciman, Chris Holmes, Mark van der Laan
NEJM AI | November 2024
-
Learning to Retrieve Iteratively for In-Context Learning
Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Ben Van Durme
EMNLP | November 2024
Outstanding Paper Award
-
Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Pass
Tong Chen, Hao Fang, Patrick Xia, Xiaodong Liu, Ben Van Durme, Luke Zettlemoyer, Jianfeng Gao, Hao Cheng
ICLR 2025 | November 2024
-
Multi-Field Adaptive Retrieval
Millicent Li, Tongfei Chen, Ben Van Durme, Patrick Xia
ICLR 2025 | October 2024
Spotlight
-
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani, Hao Fang, Patrick Xia, Eran Levy, Jacob Andreas, Ben Van Durme
IJCAI 2024 | August 2024
-
Causal Reasoning and Large Language Models: Opening a New Frontier for Causality
Emre Kiciman, Robert Osazuwa Ness, Amit Sharma, Chenhao Tan
Transactions on Machine Learning Research (TMLR) | August 2024
Outstanding Certification Finalist
Selected for presentation at ICLR 2025
-
Large Language Models as Co-Pilots for Causal Inference in Medical Studies
Ahmed Alaa, Rachael V. Phillips, Emre Kiciman, Laura B. Balzer, M. V. D. Laan, Maya Petersen
ArXiv | July 2024, 第 abs/2407.19118 卷
-
Interpreting User Requests in the Context of Natural Language Standing Instructions
Nikita Moghe, Patrick Xia, Jacob Andreas, Jason Eisner, Ben Van Durme, Harsh Jhamtani
NAACL | June 2024
-
Defending Against Indirect Prompt Injection Attacks With Spotlighting
Keegan Hines, Gary Lopez, Matthew Hall, Federico Zarfati, Yonatan Zunger, Emre Kiciman
ArXiv | March 2024, 第 abs/2403.14720 卷
-
Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models
Jingwei Yi, Yueqi Xie, Bin Benjamin Zhu, Emre Kiciman, Guangzhong Sun, Xing Xie, Fangzhao Wu
ArXiv | December 2023, 第 abs/2312.14197 卷
-
GPT-4 as an Agronomist Assistant? Answering Agriculture Exams Using Large Language Models
Bruno Silva, Leonardo Nunes, Roberto Estevão, Vijay Aski, Ranveer Chandra
October 2023
-
SCREWS: A Modular Framework for Reasoning with Revisions
Kumar Shridhar, Harsh Jhamtani, Hao Fang, Ben Van Durme, Jason Eisner, Patrick Xia
arXiv: Computation and Language | September 2023
-
The Whole Truth and Nothing But the Truth: Faithful and Controllable Dialogue Response Generation with Dataflow Transduction and Constrained Decoding
Hao Fang, Anusha Balakrishnan, Harsh Jhamtani, John Bufe, Jean Crawford, Jayant Krishnamurthy, Adam Pauls, Jason Eisner, Jacob Andreas, Dan Klein
Findings of ACL 2023 | July 2023
-
-
-
LM Agents for Coordinating Multi-User Information Gathering
Harsh Jhamtani, Jacob Andreas, Ben Van Durme
ACL | July 2025
-
Language-to-Code Translation with a Single Labeled Example
Kaj Bostrom, Harsh Jhamtani, Hao Fang, Sam Thomson, Richard Shin, Patrick Xia, Ben Van Durme, Jason Eisner, Jacob Andreas
EMNLP | November 2024
-
Ontologically Faithful Generation of Non-Player Character Dialogues
Nathaniel Weir, Ryan Thomas, Randolph D'Amore, Kellie Hill, Ben Van Durme, Harsh Jhamtani
EMNLP | November 2024
-
Learning to Retrieve Iteratively for In-Context Learning
Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Ben Van Durme
EMNLP | November 2024
Outstanding Paper Award
-
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani, Hao Fang, Patrick Xia, Eran Levy, Jacob Andreas, Ben Van Durme
IJCAI 2024 | August 2024
-
SCREWS: A Modular Framework for Reasoning with Revisions
Kumar Shridhar, Harsh Jhamtani, Hao Fang, Ben Van Durme, Jason Eisner, Patrick Xia
arXiv: Computation and Language | September 2023
-
The Whole Truth and Nothing But the Truth: Faithful and Controllable Dialogue Response Generation with Dataflow Transduction and Constrained Decoding
Hao Fang, Anusha Balakrishnan, Harsh Jhamtani, John Bufe, Jean Crawford, Jayant Krishnamurthy, Adam Pauls, Jason Eisner, Jacob Andreas, Dan Klein
Findings of ACL 2023 | July 2023
-
-
-
Multi-Field Adaptive Retrieval
Millicent Li, Tongfei Chen, Ben Van Durme, Patrick Xia
ICLR 2025 | October 2024
Spotlight
-
-
-
Artificial Intelligence–Based Copilots to Generate Causal Evidence
Maya Petersen, Ahmed Alaa, Emre Kiciman, Chris Holmes, Mark van der Laan
NEJM AI | November 2024
-
Large Language Models as Co-Pilots for Causal Inference in Medical Studies
Ahmed Alaa, Rachael V. Phillips, Emre Kiciman, Laura B. Balzer, M. V. D. Laan, Maya Petersen
ArXiv | July 2024, 第 abs/2407.19118 卷
-
-
-
Defending Against Indirect Prompt Injection Attacks With Spotlighting
Keegan Hines, Gary Lopez, Matthew Hall, Federico Zarfati, Yonatan Zunger, Emre Kiciman
ArXiv | March 2024, 第 abs/2403.14720 卷
-
-
-
Sample-Efficient Online Learning in LM Agents via Hindsight Trajectory Rewriting
Michael Y. Hu, Ben Van Durme, Jacob Andreas, Harsh Jhamtani
October 2025
-
GPT-4 as an Agronomist Assistant? Answering Agriculture Exams Using Large Language Models
Bruno Silva, Leonardo Nunes, Roberto Estevão, Vijay Aski, Ranveer Chandra
October 2023
-
-
-
LM Agents for Coordinating Multi-User Information Gathering
Harsh Jhamtani, Jacob Andreas, Ben Van Durme
ACL | July 2025
-
Direct Reasoning Optimization: LLMs Can Reward And Refine Their Own Reasoning for Open-Ended Tasks
Yifei Xu, Tusher Chakraborty, Srinagesh Sharma, Leonardo Nunes, Emre Kiciman, Songwu Lu, Ranveer Chandra
Arxiv | June 2025
-
Walk the Talk? Measuring the Faithfulness of Large Language Model Explanations
Katie Matton, Robert Osazuwa Ness, John Guttag, Emre Kiciman
2025 International Conference on Learning Representations | April 2025
-
Steering Large Language Models between Code Execution and Textual Reasoning
Yongchao Chen, Harsh Jhamtani, Srinagesh Sharma, Chuchu Fan, Chi Wang
ICLR | April 2025
-
RLTHF: Targeted Human Feedback for LLM Alignment
Yifei Xu, Tusher Chakraborty, Emre Kiciman, Bibek Aryal, Eduardo Rodrigues, Srinagesh Sharma, Roberto Estevao, Maria Angels de Luis Balaguer, Jessica Wolk, Rafael Padilha, Leonardo Nunes, Shobana Balakrishna, Songwu Lu, Ranveer Chandra
ICML’25 | February 2025
-
Language-to-Code Translation with a Single Labeled Example
Kaj Bostrom, Harsh Jhamtani, Hao Fang, Sam Thomson, Richard Shin, Patrick Xia, Ben Van Durme, Jason Eisner, Jacob Andreas
EMNLP | November 2024
-
Ontologically Faithful Generation of Non-Player Character Dialogues
Nathaniel Weir, Ryan Thomas, Randolph D'Amore, Kellie Hill, Ben Van Durme, Harsh Jhamtani
EMNLP | November 2024
-
Learning to Retrieve Iteratively for In-Context Learning
Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Ben Van Durme
EMNLP | November 2024
Outstanding Paper Award
-
Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Pass
Tong Chen, Hao Fang, Patrick Xia, Xiaodong Liu, Ben Van Durme, Luke Zettlemoyer, Jianfeng Gao, Hao Cheng
ICLR 2025 | November 2024
-
Multi-Field Adaptive Retrieval
Millicent Li, Tongfei Chen, Ben Van Durme, Patrick Xia
ICLR 2025 | October 2024
Spotlight
-
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani, Hao Fang, Patrick Xia, Eran Levy, Jacob Andreas, Ben Van Durme
IJCAI 2024 | August 2024
-
Interpreting User Requests in the Context of Natural Language Standing Instructions
Nikita Moghe, Patrick Xia, Jacob Andreas, Jason Eisner, Ben Van Durme, Harsh Jhamtani
NAACL | June 2024
-
The Whole Truth and Nothing But the Truth: Faithful and Controllable Dialogue Response Generation with Dataflow Transduction and Constrained Decoding
Hao Fang, Anusha Balakrishnan, Harsh Jhamtani, John Bufe, Jean Crawford, Jayant Krishnamurthy, Adam Pauls, Jason Eisner, Jacob Andreas, Dan Klein
Findings of ACL 2023 | July 2023
-
-
-
Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots
Margaret Capetz, Swati Sharma, Rafael Padilha, Peder Olsen, Jessica Wolk, Emre Kiciman, Ranveer Chandra
ArXiv | November 2024, 第 abs/2411.16872 卷
-
Artificial Intelligence–Based Copilots to Generate Causal Evidence
Maya Petersen, Ahmed Alaa, Emre Kiciman, Chris Holmes, Mark van der Laan
NEJM AI | November 2024
-
Causal Reasoning and Large Language Models: Opening a New Frontier for Causality
Emre Kiciman, Robert Osazuwa Ness, Amit Sharma, Chenhao Tan
Transactions on Machine Learning Research (TMLR) | August 2024
Outstanding Certification Finalist
Selected for presentation at ICLR 2025
-
Large Language Models as Co-Pilots for Causal Inference in Medical Studies
Ahmed Alaa, Rachael V. Phillips, Emre Kiciman, Laura B. Balzer, M. V. D. Laan, Maya Petersen
ArXiv | July 2024, 第 abs/2407.19118 卷
-
Defending Against Indirect Prompt Injection Attacks With Spotlighting
Keegan Hines, Gary Lopez, Matthew Hall, Federico Zarfati, Yonatan Zunger, Emre Kiciman
ArXiv | March 2024, 第 abs/2403.14720 卷
-
Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models
Jingwei Yi, Yueqi Xie, Bin Benjamin Zhu, Emre Kiciman, Guangzhong Sun, Xing Xie, Fangzhao Wu
ArXiv | December 2023, 第 abs/2312.14197 卷
-
SCREWS: A Modular Framework for Reasoning with Revisions
Kumar Shridhar, Harsh Jhamtani, Hao Fang, Ben Van Durme, Jason Eisner, Patrick Xia
arXiv: Computation and Language | September 2023
-