Entropy minimization (EM) trains the model to concentrate even more probability mass on its most confident outputs. We show that this simple objective alone, without any labeled data, can substantially improve large language models' (LLMs) performance on challenging math, physics, and coding tasks. We explore three approaches: (1) EM-FT minimizes token-level entropy similarly to instruction finetuning, but on unlabeled outputs drawn from the model; (2) EM-RL: reinforcement learning with negative entropy as the only reward to maximize; (3) EM-INF: inference-time logit adjustment to reduce entropy without any training data or parameter updates. On Qwen-7B, EM-RL, without any labeled data, achieves comparable or better performance than strong RL baselines such as GRPO and RLOO that are trained on 60K labeled examples. Furthermore, EM-INF enables Qwen-32B to match or exceed the performance of proprietary models like GPT-4o, Claude 3 Opus, and Gemini 1.5 Pro on the challenging SciCode benchmark, while being 3x more efficient than self-consistency and sequential refinement. Our findings reveal that many pretrained LLMs possess previously underappreciated reasoning capabilities that can be effectively elicited through entropy minimization alone, without any labeled data or even any parameter updates.
Twitter（推特）；Imperial College London（伦敦帝国理工学院）；University of Oxford（牛津大学）
While Graph Neural Networks (GNNs) have recently become the de facto standard for modeling relational data, they impose a strong assumption on the availability of the node or edge features of the graph. In many real-world applications, however, features are only partially available; for example, in social networks, age and gender are available only for a small subset of users. We present a general approach for handling missing features in graph machine learning applications that is based on minimization of the Dirichlet energy and leads to a diffusion-type differential equation on the graph. T…
Deep learning has excelled on complex pattern recognition tasks such as image classification and object recognition. However, it struggles with tasks requiring nontrivial reasoning, such as algorithmic computation. Humans are able to solve such tasks through iterative reasoning -- spending more time thinking about harder tasks. Most existing neural networks, however, exhibit a fixed computational budget controlled by the neural network architecture, preventing additional computational processing on harder tasks. In this work, we present a new framework for iterative reasoning with neural netwo…
Solving Challenging Math Word Problems Using GPT-4 Code Interpreter with Code-based Self-Verification
The Chinese University of Hong Kong（香港中文大学）；Nanjing University（南京大学）；University of Science and Technology of China（中国科学技术大学）；Tsinghua University（清华大学）；City University of Hong Kong（香港城市大学）；Changsha University of Science and Technology（长沙理工大学）；Tufts University（塔夫茨大学）
Spectral Logit Sculpting: Adaptive Low-Rank Logit Transformation for Controlled Text Generation
Zhejiang University（浙江大学）；Binjiang Institute of Zhejiang University（浙江大学滨江学院）；Hangzhou Dianzi University（杭州电子科技大学）；People’s Public Security University of China（中国人民公安大学）；Guangzhou University（广州大学）
DS-Agent: Automated Data Science by Empowering Large Language Models with Case-Based Reasoning
School of Artificial Intelligence, Jilin University（吉林大学人工智能学院）；Engineering Research Center of Knowledge-Driven Human-Machine Intelligence, Jilin University（吉林大学知识驱动人机智能工程研究中心）；International Center of Future Science, Jilin University（吉林大学未来科学国际合作联合实验室）；Shanghai Jiao Tong University（上海交通大学）；University College London（伦敦大学学院）
SPOT-E: Test-Time Entropy Shaping with Visual Spotlights for Frozen VLMs
National University of Singapore（新加坡国立大学）；Fudan University（复旦大学）；Technical University of Munich（慕尼黑工业大学）；Sagenic Tech；Zhejiang University（浙江大学）；vivo；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）
