潜在思维流：大型语言模型中的高效潜在推理
Latent Thought Flow: Efficient Latent Reasoning in Large Language Models
AI总结：
提出Latent Thought Flow (LTF)方法，将推理建模为可变长度连续轨迹，通过连续GFlowNet训练采样器匹配奖励后验，在提升准确率9.5%的同时平均减少推理长度27.2%。
摘要：
大型语言模型（LLMs）越来越依赖中间推理，然而显式的思维链（CoT）存在语言空间瓶颈：每个思维必须解码为token，导致高推理开销。潜在推理将思考过程转移到连续空间，但现有方法大多学习确定性或奖励最大化路径，缺乏在具有不同正确性和成本的轨迹间分配概率的原则性方法。我们提出潜在思维流（LTF），将推理建模为可变长度连续轨迹，并训练采样器以匹配由答案质量和计算成本定义的奖励诱导后验。我们使用具有随机潜在转移的连续GFlowNet实例化该方法。为处理稀疏答案监督，我们引入熵加权子轨迹平衡目标以获取中间奖励，以及参考先验正则化器以锚定探索。在微调和迁移学习设置下的实验表明，与强潜在推理基线相比，LTF在平均减少推理长度27.2%的同时，准确率提升9.5%，优于显式CoT和潜在推理基线。
英文摘要：
Large Language Models (LLMs) increasingly rely on intermediate reasoning, yet explicit Chain-of-Thought (CoT) suffers from a linguistic space bottleneck: each thought must be decoded into tokens, causing high inference overhead. Latent reasoning moves deliberation into continuous space, but existing methods mostly learn deterministic or reward-maximizing paths, lacking a principled way to allocate probability across trajectories with different correctness and costs. We propose Latent Thought Flow (LTF), which models reasoning as variable-length continuous trajectories and trains a sampler to match a reward-induced posterior over answer quality and computation cost. We instantiate this with a continuous GFlowNet using stochastic latent transitions. To handle sparse answer supervision, we introduce an Entropy-Weighted Subtrajectory Balance objective for intermediate rewards and a reference-prior regularizer to anchor exploration. Experiments under finetuning and transfer learning settings show that LTF outperforms explicit CoT and latent reasoning baselines, improving accuracy by 9.5% while reducing reasoning length by 27.2% on average compared with strong latent reasoning baselines.
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。
