One-shot Entropy Minimization for Language Model Reasoning

In this work, we propose One-shot Entropy Minimization (EM), a simple and fully unsupervised post-training approach that significantly improves reasoning and generation performance using only a single unlabeled data and approximately ten gradient steps. To avoid data contamination, we pretrain a 7-billion-parameter language model from scratch with strictly decontaminated data. Despite its extreme simplicity, one-shot EM yields substantial performance gains and improves reasoning abilities across a broad range of domains, including mathematical reasoning, logical reasoning, and coding. We further show that entropy minimization induces a characteristic right-skewed logit shift, amplifying high-probability tokens while suppressing low-probability tails, in contrast to reinforcement learning. Our findings suggest that entropy minimization primarily acts as a distribution shaping mechanism rather than a conventional learning process, offering an efficient and practical algorithm for post-training large language models.

We propose a simple post-training method for language models that uses only one unlabeled example and about 10 training steps to significantly improve reasoning ability, achieving performance comparable to reinforcement learning with tens of thousands of steps.