通过递归激活关键层，让模型推理更高效。
Encode, Think, Decode: Scaling test-time reasoning with recursive latent thoughts
- 在训练中期迭代特定推理层，增强潜在思维能力。
- 在GSM8K上提升28.4%准确率，MATH上提升36%。
- 无需改架构或参数，适合资源受限场景部署。
现有提升大语言模型推理能力的方法主要依赖增加参数量或训练数据规模，或在推理时生成复杂思维链。受可解释性研究启发——推理关键计算集中在少数层中，本文提出编码-思考-解码（ETD）方法：在训练中期训练模型反复迭代一小部分与推理相关的层。该方法在保持原始架构、参数量、超参数和训练数据不变的前提下，放大潜在推理能力。在推理时对选定层进行递归操作，ETD模型在17个推理基准上均取得显著提升，使用OLMo-2 1B Base模型时，在GSM8K上相对准确率提高28.4%，在MATH上提升36%。我们还探索了自适应深度策略，按输入词元动态调整计算量。结果表明，递归潜层推理为强化大模型推理提供了一条简单而有效的新路径。
原文摘要 · Abstract (English)
Most efforts to improve the reasoning capabilities of large language models (LLMs) involve either scaling the number of parameters and the size of training data, or scaling inference computation by letting models generate complex chains of thought. Motivated by interpretability studies showing that the crucial computation required for reasoning tasks is concentrated in a limited range of layers, we introduce Encode-Think-Decode (ETD), a method that enhances the reasoning capabilities of a base model by training it to iterate over a small subset of reasoning-relevant layers during the mid-training stage. ETD amplifies latent reasoning while preserving the original architecture, parameter count, hyperparameters, and training data composition. When iterating on the selected layers at inference time, ETD models yield substantial gains on 17 reasoning benchmarks, including +28.4% relative accuracy improvement on GSM8K and +36% on MATH with the OLMo-2 1B Base model. We also explore an adaptive depth strategy that adjusts the computation per input token. Our results show that recursive latent reasoning offers a simple and effective path to stronger LLM reasoning.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。