{"slug": "what-fits-into-few-tokens-doesn-t-overfit", "title": "What Fits (Into Few Tokens) Doesn't Overfit", "summary": "A June 9, 2026 arXiv paper finds that LLM-driven ML research agents reproduce and discover high-performance models even when constrained to extremely short prompts or one-bit feedback, across 8 datasets spanning tabular classification, vision, language modeling, diffusion modeling, and reward modeling. The authors attribute the lack of overfitting in benchmark-driven ML to a description-length effect: successful strategies occupy a low-complexity region of strategy space. When validation-set overfitting was deliberately induced, the results failed to reproduce with short prompts, supporting the hypothesis.", "body_md": "# Computer Science > Artificial Intelligence\n\n  [Submitted on 9 Jun 2026]\n\n# Title:What Fits (Into Few Tokens) Doesn't Overfit: Compression and Generalization in ML Research Agents\n\n[View PDF](https://arxiv.org/pdf/2606.11045)\n\n[HTML (experimental)](https://arxiv.org/html/2606.11045v1)\n\nAbstract:Reusing a held-out benchmark adaptively should, in principle, invite overfitting. Yet benchmark-driven machine learning (ML) has produced surprisingly little overfitting in practice. An attractive hypothesis is that successful ML strategies are highly compressible. We study this in the setting of LLM-driven research agents, where the hypothesis becomes directly testable via two complementary information bottlenecks. In \\emph{output compression}, an exploration agent adaptively searches for high-performance models using a validation set, and we test whether a fresh ``reproducer agent'' can reproduce its performance given only an extremely short prompt and the training data. In \\emph{input compression}, the explorer receives only one-bit feedback indicating whether each submitted model improves on the running best. Across 8 datasets spanning tabular classification, vision, language modeling, diffusion modeling, and reward modeling, we find that these bottlenecks have little effect on performance: short prompts and compressible feedback are sufficient to reproduce and find high-performance models. The hypothesis is falsifiable: when we deliberately induce validation-set overfitting, the results fail to reproduce with short prompts. Taken together, our results support a description-length explanation for the lack of overfitting in benchmark-driven ML: successful strategies occupy a low-complexity region of strategy space.\n    \n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer \n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers \n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps \n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations \n\n*(*[What are Smart Citations?](https://www.scite.ai/))\n# Code, Data and Media Associated with this Article\n\nalphaXiv \n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers \n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub \n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub \n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face \n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast \n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))\n# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower \n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender \n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\n# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [**Learn more about arXivLabs**](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/what-fits-into-few-tokens-doesn-t-overfit", "canonical_source": "https://arxiv.org/abs/2606.11045", "published_at": "2026-09-17 11:34:47+00:00", "updated_at": "2026-09-17 11:56:41.798065+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research", "ai-agents"], "entities": ["arXiv", "LLM-driven research agents", "reproducer agent", "exploration agent"], "alternates": {"html": "https://wpnews.pro/news/what-fits-into-few-tokens-doesn-t-overfit", "markdown": "https://wpnews.pro/news/what-fits-into-few-tokens-doesn-t-overfit.md", "text": "https://wpnews.pro/news/what-fits-into-few-tokens-doesn-t-overfit.txt", "jsonld": "https://wpnews.pro/news/what-fits-into-few-tokens-doesn-t-overfit.jsonld"}}