Entropy-informed Decoding: Adaptive Information-Driven Branching

Large language models (LLMs) achieve remarkable generative performance, yet their output quality is dependent on the decoding strategy. While sampling-based methods (e.g., top-k, nucleus) and search-and-select based methods (e.g., beam search, best-of-n, majority voting) can improve upon greedy decoding, both approaches suffer from limitations: sampling generally commits to a single path, while search often expends excessive computation regardless of task complexity. To address these, we introduce Entropy-informed DEcodiNg (EDEN), a plug-and-play, model-agnostic decoding framework that adaptively allocates computation based on the model’s own uncertainty, approximating higher-width beam search with fewer expansions. At each generation step, EDEN estimates the entropy of the output token distribution and adjusts the branching factor monotonically with the entropy, expanding more candidates in high-entropy regions and following a greedier path in low-entropy regions, improving token efficiency. Experiments across complex tasks, including mathematical reasoning, code generation, and scientific questions, demonstrate that EDEN consistently improves output quality over existing decoding strategies, achieving better accuracy-expansion trade-offs than fixed-width beam search. By treating next-token selection as a noisy maximisation problem, we prove that branching factors monotone in entropy are guaranteed to find better (i.e. more probable) continuations than any fixed branching factor within the same total expansion budget, and derive explicit regret rates characterising the benefit of the adaptive allocation.

People do not think equally hard about every word in a sentence or every step in a problem. We move quickly when the answer is obvious, and slow down when we reach a difficult choice. EDEN brings this idea to language models. Instead of searching the same amount at every step, EDEN lets the model search more when it is unsure and search less when it is confident. This turns decoding into a form of adaptive thinking: easy steps are handled quickly, while tricky decision points get extra focus. In tasks like math, coding, and science questions, this helped the model find better answers while exploring fewer possibilities than standard search methods. The result is a simple way for language models to spend their effort more like people do: not everywhere, but where it matters most.