Large language models (LLMs) are increasingly used to automate AI research, including language-model (LM) development, offering a potential path toward recursive self-improvement. However, turning experimental feedback into a learned ability to propose effective improvements across models faces two challenges. First, training such an ability with real feedback is expensive: reinforcement learning (RL) requires evaluating tens of thousands of proposals, each potentially requiring a full model-training run. Even at ten GPU-hours per run, reward collection alone would consume hundreds of thousands of GPU-hours, motivating a cheaper surrogate. Second, surrogate feedback can become unreliable as policy optimization favors overestimated proposals when the proposal distribution shifts. To address these challenges, we propose LM2LM, which trains LLMs with RL to generate effective LM improvement proposals. To provide scalable feedback, we use a research world model (RWM), a frozen LLM that predicts proposal gains from real experimental records provided in context, as the reward model for RL. To maintain feedback reliability, we periodically execute current-policy proposals ranked highly by the RWM and add their measured outcomes to its context. Across eight autoregressive and diffusion LM development environments, real-execution evaluations on held-out tasks show that LM2LM enables Qwen3.8-27B to outperform GPT-6-sol, even when GPT-6-sol is given 300 real experimental records in context, while requiring 33% lower data-generation cost. LM2LM also generalizes to a model family unseen during training.
Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP
Center for Information and Language Processing (CIS), LMU Munich(慕尼黑大学信息与语言处理中心);Institute of Social and Cultural Anthropology, LMU Munich(慕尼黑大学社会与文化人类学研究所)
PatientDx: Merging Large Language Models for Protecting Data-Privacy in Healthcare
Université de Toulouse(图卢兹大学);IRIT UMR 5505;Centre Hospitalier Universitaire de Toulouse(图卢兹大学医院中心);INSERM(法国国家健康与医学研究院);Faculté de Médecine Université de Toulouse(图卢兹大学医学院)
PAPERCLIP: Associating Astronomical Observations and Natural Language with Multi-Modal Models
The NSF AI Institute for Artificial Intelligence and Fundamental Interactions(美国国家科学基金会人工智能与基础交互研究所);Center for Theoretical Physics, Massachusetts Institute of Technology(麻省理工学院理论物理中心);Department of Physics, Harvard University(哈佛大学物理系);Department of Physics, Massachusetts Institute of Technology(麻省理工学院物理系)
Teaching Programming in the Age of Generative AI: Insights from Literature, Pedagogical Proposals, and Student Perspectives
Departamento de Sistemas de Información - Universidad del Bío-Bío(信息系统系 - 生物-生物大学);Escuela de Informática y Telecomunicaciones - DuocUC(信息与电信工程学院 - DuocUC)