Continually Improving AI

We aim to build AI systems that learn from feedback, experience, and their own failures to become more capable over time.

Related publications

2026

  1. ImProver 2: Iteratively Self-Improving LMs for Neurosymbolic Proof Optimization
    ImProver 2: Iteratively Self-Improving LMs for Neurosymbolic Proof Optimization
    Riyaz Ahuja, Tate Rowney, Jeremy Avigad, and 1 more author
    arXiv, 2026
  2. AdaExplore: Failure-Driven Adaptation and Diversity-Preserving Search for Efficient Kernel Generation
    AdaExplore: Failure-Driven Adaptation and Diversity-Preserving Search for Efficient Kernel Generation
    Weihua Du, Jingming Zhuo, Yixin Dong, and 9 more authors
    arXiv, 2026
    Code Website
  3. Propose, Solve, Verify: Self-Play Through Formal Verification
    Propose, Solve, Verify: Self-Play Through Formal Verification
    Alex Wilf, Pranjal Aggarwal, Bryan Parno, and 4 more authors
    2026
    ICML 2026
    Code
  4. RefineBench: Evaluating Refinement Capability of Language Models via Checklists
    RefineBench: Evaluating Refinement Capability of Language Models via Checklists
    Young-Jun Lee, Seungone Kim, Byung-Kwan Lee, and 6 more authors
    2026
    ICLR 2026

2025

  1. Rewarding the Unlikely: Lifting GRPO Beyond Distribution Sharpening
    Rewarding the Unlikely: Lifting GRPO Beyond Distribution Sharpening
    Andre He, Daniel Fried, and Sean Welleck
    2025
    EMNLP 2025 Oral
    Code

2024

  1. Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision
    Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision
    Zhiqing Sun, Longhui Yu, Yikang Shen, and 4 more authors
    In Advances in Neural Information Processing Systems, 2024

2023

  1. Self-Refine: Iterative Refinement with Self-Feedback
    Self-Refine: Iterative Refinement with Self-Feedback
    Aman Madaan, Niket Tandon, Prakhar Gupta, and 13 more authors
    In Thirty-seventh Conference on Neural Information Processing Systems, 2023
  2. Generating Sequences by Learning to Self-Correct
    Generating Sequences by Learning to Self-Correct
    Sean Welleck, Ximing Lu, Peter West, and 4 more authors
    In The Eleventh International Conference on Learning Representations , 2023