CoLT：基于链式潜在工具调用的推理
CoLT: Reasoning with Chain of Latent Tool Calls
AI总结：
CoLT通过将潜在推理转化为工具调用，实现高效推理，提升模型效率与准确性，适用于多种解码器结构和强化学习算法。
摘要：
链式思维（CoT）是增强大型语言模型（LLMs）推理能力的关键技术，而潜在推理方法已被提出以加速低效的token级推理链。我们注意到现有的潜在推理方法通常需要模型结构增强和穷尽训练，限制了其更广泛的应用。在本文中，我们提出CoLT，一种新的框架，将潜在推理实现为『工具调用』。与完全在潜在空间中推理不同，CoLT生成包含推理步骤信息的种子token。当触发潜在工具调用时，较小的外部模型将种子token的隐藏状态作为输入，并将种子token解包回完整的推理步骤。这样，我们可以在显式token空间中让主模型进行推理，保持其能力的同时提高效率。在四个数学数据集上的实验结果表明，CoLT在准确率和推理长度方面均优于基线潜在模型，并且能够与强化学习算法和不同的解码器结构兼容。
英文摘要：
Chain-of-Thought (CoT) is a critical technique in enhancing the reasoning ability of Large Language Models (LLMs), and latent reasoning methods have been proposed to accelerate the inefficient token-level reasoning chain. We notice that existing latent reasoning methods generally require model structure augmentation and exhaustive training, limiting their broader applicability. In this paper, we propose CoLT, a novel framework that implements latent reasoning as ``tool calls''. Instead of reasoning entirely in the latent space, CoLT generates seed tokens that contain information of a reasoning step. When a latent tool call is triggered, a smaller external model will take the hidden states of seed tokens as its input, and unpack the seed tokens back to a full reasoning step. In this way, we can ensure that the main model reasons in the explicit token space, preserving its ability while improving efficiency. Experimental results on four mathematical datasets demonstrate that CoLT achieves higher accuracy and shorter reasoning length than baseline latent models, and is compatible with reinforcement learning algorithms and different decoder structures.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
