Despite their extensive pretraining knowledge and strong reasoning capabilities, large language models (LLMs) struggle to adaptively gather information in experimental design tasks. We study how to am
Large language models learn adaptive information gathering by training reasoning with information-gain rewards, improving experimental design efficiency and generalizing across tasks.
Despite their extensive pretraining knowledge and strong reasoning capabilities, large language models (LLMs) struggle to adaptively gather information in experimental design tasks. We study how to amortize sequential Bayesian experimental design (BED) into an LLM by explicitly training the reasoning that guides action selection using information-gain rewards. Our approach specifies the experimental generative model independently of the LLM's own beliefs, allowing us to optimize expected information gain under a model of our choosing. We investigate the design choices for amortization, including alternative generative-model constructions, reward estimation, hypothesis-space specification,…
Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.
Amortizing Sequential Experimental Design in Reasoning LLMs was first seen on Hugging Face · Daily papers on Oct 29, 2026. Since then it has appeared in 31 checks on 5 different days. Its best position was #24 of 30, reached on Oct 30, 05:28 UTC. It dropped off the list after Nov 2, 01:28 UTC.
Position on Hugging Face · Daily papers at every check (lower is better).
Best rank per day (UTC).
Platform metric as reported by Hugging Face.
Archived daily, weekly and monthly lists that include this entry.
Not at the moment. It was last on the list on Nov 2, 01:28 UTC, after 5 days of trending.
#24 on Hugging Face · Daily papers, on Oct 30, 05:28 UTC. Trendoday checks this list every 120 minutes.
It was first seen on Oct 29, 2026 and has been on the list on 5 days (31 checks) since.