BALAR : A Bayesian Agentic Loop for Active Reasoning
For LLM-based interactive systems, BALAR provides a principled, task-agnostic method to actively gather missing information without fine-tuning.
BALAR introduces a Bayesian active reasoning loop for LLMs that selects clarifying questions via mutual information maximization, achieving 14.6%, 38.5%, and 30.5% higher accuracy on three diverse benchmarks compared to baselines.
Large language models increasingly operate in interactive settings where solving a task requires multiple rounds of information exchange with a user. However, most current systems treat dialogue reactively and lack a principled mechanism to reason about what information is missing and which question should be asked next. We propose BALAR (Bayesian Agentic Loop for Active Reasoning), a task-agnostic outer-loop algorithm that requires no fine-tuning and enables structured multi-turn interaction between an LLM agent and a user. BALAR maintains a structured belief over latent states, selects clarifying questions by maximizing expected mutual information, and dynamically expands its state representation when the current one proves insufficient. We evaluate BALAR on three diverse benchmarks: AR-Bench-DC (detective cases), AR-Bench-SP (thinking puzzles), and iCraft-MD (clinical diagnosis). BALAR significantly outperforms all baselines across all three benchmarks, with $14.6\%$ higher accuracy on AR-Bench-DC, $38.5\%$ on AR-Bench-SP, and $30.5\%$ on iCraft-MD.