3 citations · 13 across the 21 of their papers we have counts for
24 papers
Don't Forget What You Just Learned: Self-Directed Repair Turns Agent Mistakes Into Lasting Lessons
Yuval Kansal, Karen Ullrich, Polina Kirichenko +4
Intelligent agents should learn from their mistakes so that experience leads to continuous improvement. Yet today's agents recover from a mistake only temporarily: in-context refle…
Agent Plasticity: Measuring Self-Improvement Through Experience
Harman Singh, Anton Bakhtin, Rulin Shao +8
AI agents increasingly operate in environments where they can diagnose failures and improve through experience, yet existing evaluations largely measure what an agent can do at a f…
Autodata: An agentic data scientist to create high quality synthetic data
Ilia Kulikov, Chenxi Whitehouse, Tianhao Wu +12
We introduce Autodata, a general method that enables AI agents to act as data scientists who build high quality training and evaluation data. We show how to train (meta-optimize) s…
Reasoning over mathematical objects: on-policy reward modeling and test time aggregation
Pranjal Aggarwal, Marjan Ghazvininejad, Seungone Kim +18
The ability to precisely derive mathematical objects is a core requirement for downstream STEM applications, including mathematics, physics, and chemistry, where reasoning must cul…
Safety Alignment of LMs via Non-cooperative Games
Anselm Paulus, Ilia Kulikov, Brandon Amos +4
Ensuring the safety of language models (LMs) while maintaining their usefulness remains a critical challenge in AI alignment. Current approaches rely on sequential adversarial trai…
SPICE: Self-Play In Corpus Environments Improves Reasoning
Bo Liu, Chuanyang Jin, Seungone Kim +7
Self-improving systems require environmental interaction for continuous adaptation. We introduce SPICE (Self-Play In Corpus Environments), a reinforcement learning framework where…
Hybrid Reinforcement: When Reward Is Sparse, It's Better to Be Dense
Leitian Tao, Ilia Kulikov, Swarnadeep Saha +5
Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable,…
The Majority is not always right: RL training for solution aggregation
Wenting Zhao, Pranjal Aggarwal, Swarnadeep Saha +3
Scaling up test-time compute, by generating multiple independent solutions and selecting or aggregating among them, has become a central paradigm for improving large language model…
OptimalThinkingBench: Evaluating Over and Underthinking in LLMs
Pranjal Aggarwal, Seungone Kim, Jack Lanchantin +4
Thinking LLMs solve complex tasks at the expense of increased compute and overthinking on simpler problems, while non-thinking LLMs are faster and cheaper but underthink on harder…
Learning to Reason for Factuality
Xilun Chen, Ilia Kulikov, Vincent-Pierre Berges +5
Reasoning Large Language Models (R-LLMs) have significantly advanced complex reasoning tasks but often struggle with factuality, generating substantially more hallucinations than t…
NaturalThoughts: Selecting and Distilling Reasoning Traces for General Reasoning Tasks
Yang Li, Youssef Emad, Karthik Padthe +8
Recent work has shown that distilling reasoning traces from a larger teacher model via supervised finetuning outperforms reinforcement learning with the smaller student model alone…
CoT-Self-Instruct: Building high-quality synthetic prompts for reasoning and non-reasoning tasks
Ping Yu, Jack Lanchantin, Tianlu Wang +6
We propose CoT-Self-Instruct, a synthetic data generation method that instructs LLMs to first reason and plan via Chain-of-Thought (CoT) based on given seed tasks, and then generat…
Bridging Offline and Online Reinforcement Learning for LLMs
Jack Lanchantin, Angelica Chen, Janice Lan +9
We investigate the effectiveness of reinforcement learning methods for finetuning large language models when transitioning from offline to semi-online to fully online regimes for b…
J1: Incentivizing Thinking in LLM-as-a-Judge via Reinforcement Learning
Chenxi Whitehouse, Tianlu Wang, Ping Yu +4
The progress of AI is bottlenecked by the quality of evaluation, making powerful LLM-as-a-Judge models a core solution. The efficacy of these judges depends on their chain-of-thoug…
Post-training an LLM for RAG? Train on Self-Generated Demonstrations
Matthew Finlayson, Ilia Kulikov, Daniel M. Bikel +3
Large language models (LLMs) often struggle with knowledge intensive NLP tasks, such as answering "Who won the latest World Cup?" because the knowledge they learn during training m…
LLM Pretraining with Continuous Concepts
Jihoon Tack, Jack Lanchantin, Jane Yu +7
Next token prediction has been the standard training objective used in large language model pretraining. Representations are learned as a result of optimizing for token-level perpl…
Learning to Solve and Verify: A Self-Play Framework for Code and Test Generation
Zi Lin, Sheng Shen, Ilia Kulikov +3
Recent advances in large language models (LLMs) have improved their performance on coding benchmarks. However, improvement is plateauing due to the exhaustion of readily available…
NaturalReasoning: Reasoning in the Wild with 2.8M Challenging Questions
Weizhe Yuan, Jane Yu, Song Jiang +8
Scaling reasoning capabilities beyond traditional domains such as math and coding is hindered by the lack of diverse and high-quality questions. To overcome this limitation, we int…
Diverse Preference Optimization
Jack Lanchantin, Angelica Chen, Shehzaad Dhuliawala +4
Post-training of language models, either through reinforcement learning, preference optimization or supervised finetuning, tends to sharpen the output probability distribution and…
Adaptive Decoding via Latent Preference Optimization
Shehzaad Dhuliawala, Ilia Kulikov, Ping Yu +4
During language model decoding, it is known that using higher temperature sampling gives more creative responses, while lower temperatures are more factually accurate. However, suc…
Self-Taught Evaluators
Tianlu Wang, Ilia Kulikov, Olga Golovneva +7
Model-based evaluation is at the heart of successful model development -- as a reward model for training, and as a replacement for human evaluation. To train such evaluators, the s…
Distilling System 2 into System 1
Ping Yu, Jing Xu, Jason Weston +1
Large language models (LLMs) can spend extra compute during inference to generate intermediate thoughts, which helps to produce better final responses. Since Chain-of-Thought (Wei…
Following Length Constraints in Instructions
Weizhe Yuan, Ilia Kulikov, Ping Yu +4
Aligned instruction following models can better fulfill user requests than their unaligned counterparts. However, it has been shown that there is a length bias in evaluation of suc…
From Decoding to Meta-Generation: Inference-time Algorithms for Large Language Models
Sean Welleck, Amanda Bertsch, Matthew Finlayson +5
One of the most striking findings in modern research on large language models (LLMs) is that scaling up compute during training leads to better results. However, less attention has…