Jonas Hübotter

Publications

* denotes equal contribution.

2026

  1. Aligning Language Models from User Interactions
    Thomas Kleine Buening, Jonas Hübotter, Barna Pásztor, Idan Shenfeld, Giorgia Ramponi, and Andreas Krause
    Conference on Language Modeling (2026)
    Oral Presentation at ICML 2026 Workshop on Continual Adaptation at Scale.
  2. Majority Voting for Code Generation
    Tim Launer, Jonas Hübotter, Marco Bagatella, Ido Hakimi, and Andreas Krause
    Workshop on Test-Time Updates @ International Conference on Learning Representations (2026)
  3. Tool-R0: Self-Evolving LLM Agents for Tool-Learning from Zero Data
    Emre Can Acikgoz, Chen Qian, Jonas Hübotter, Heng Ji, Dilek Hakkani-Tür, and Gokhan Tur
    arXiv preprint arXiv:2602.21320
  4. Reinforcement Learning via Self-Distillation
    Jonas Hübotter, Frederike Lübeck*, Lejs Behric*, Anton Baumann*, Marco Bagatella, Daniel Marta, Ido Hakimi, Idan Shenfeld, Thomas Kleine Buening, Carlos Guestrin, and Andreas Krause
    International Conference on Machine Learning (2026)
    Best Paper Award at ICLR 2026 Workshop on Test-Time Updates and Oral Presentation at ICLR 2026 Workshop on Scaling Post-training for LLMs.
  5. Self-Distillation Enables Continual Learning
    Idan Shenfeld, Mehul Damani, Jonas Hübotter, and Pulkit Agrawal
    International Conference on Machine Learning (2026)
    Best Paper Award at ICLR 2026 Workshop on Lifelong Agents.
  6. Test-time Offline Reinforcement Learning on Goal-related Experience
    Marco Bagatella*, Mert Albaba*, Jonas Hübotter, Georg Martius, and Andreas Krause
    International Conference on Machine Learning (2026)

2025

  1. Learning on the Job: Test-Time Curricula for Targeted Reinforcement Learning
    Jonas Hübotter*, Leander Diaz-Bone*, Ido Hakimi, Andreas Krause, and Moritz Hardt
    arXiv preprint arXiv:2510.04786
  2. Specialization after Generalization: Towards Understanding Test-Time Training in Foundation Models
    Jonas Hübotter*, Patrik Wolf*, Alexander Shevchenko*, Dennis Jüni, Andreas Krause, and Gil Kur
    International Conference on Learning Representations (2026)
    Oral Presentation at NeurIPS 2025 Workshop on Continual and Compatible Foundation Model Updates.
  3. Maximizing Prefix-Confidence at Test-Time Efficiently Improves Mathematical Reasoning
    Matthias Otth, Jonas Hübotter, Ido Hakimi, and Andreas Krause
    Workshop on Test-Time Scaling and Reasoning Models @ Conference on Language Modeling (2025)
  4. DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning
    Leander Diaz-Bone*, Marco Bagatella*, Jonas Hübotter*, and Andreas Krause
    Advances in Neural Information Processing Systems (2025)
  5. Local Mixtures of Experts: Essentially Free Test-Time Training via Model Merging
    Ryo Bertolissi*, Jonas Hübotter*, Ido Hakimi, and Andreas Krause
    Conference on Language Modeling (2025)
  6. Probabilistic Artificial Intelligence
    Andreas Krause and Jonas Hübotter
    arXiv preprint arXiv:2502.05244
  7. LITE: Efficiently Estimating Gaussian Probability of Maximality
    Nicolas Menet, Jonas Hübotter, Parnian Kassraie, and Andreas Krause
    International Conference on Artificial Intelligence and Statistics (2025)

2024

  1. Active Fine-Tuning of Multi-Task Policies
    Marco Bagatella, Jonas Hübotter, Georg Martius, and Andreas Krause
    International Conference on Machine Learning (2025)
  2. Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs
    Jonas Hübotter, Sascha Bongni, Ido Hakimi, and Andreas Krause
    International Conference on Learning Representations (2025)
    Best Paper Award at NeurIPS 2024 Workshop on Fine-Tuning in Modern Machine Learning.
  3. Transductive Active Learning: Theory and Applications
    Jonas Hübotter, Bhavya Sukhija, Lenart Treven, Yarden As, and Andreas Krause
    Advances in Neural Information Processing Systems (2024)
    Oral Presentation at ICML 2024 Workshop on Aligning Reinforcement Learning Experimentalists and Theorists.

2023

  1. Efficient Exploration in Continuous-time Model-based Reinforcement Learning
    Lenart Treven, Jonas Hübotter, Bhavya Sukhija, Florian Dörfler, and Andreas Krause
    Advances in Neural Information Processing Systems (2023)
  2. Tuning Legged Locomotion Controllers via Safe Bayesian Optimization
    Daniel Widmer, Dongho Kang, Bhavya Sukhija, Jonas Hübotter, Andreas Krause, and Stelian Coros
    Conference on Robot Learning (2023)

2022

  1. A Cut-Matching Game for Constant-Hop Expanders
    Bernhard Haeupler, Jonas Huebotter, and Mohsen Ghaffari
    ACM-SIAM Symposium on Discrete Algorithms (2025)