arXiv cs.AI· Seokhyun Lee, Seongjun Lee, Changhee Lee·· 13 小时前AI 评分29
PLRM：可并行化的概率潜在推理模型，建模多样化推理轨迹
Parallelizable Probabilistic Latent Reasoning for Diverse Reasoning Trajectories
AI 导读
研究者提出概率潜在推理模型 PLRM，一个可并行的变分框架，用 JEPA 式目标让后验从观测推理轨迹中并行推断潜在分布、先验通过 teacher forcing 预测后续潜在分布。PLRM 引入离散策略变量与混合先验，保留多条推理轨迹形成的高斯混合结构而非坍缩为单一高斯。在单轨迹数据集上其表现与确定性潜在推理模型相当，在多轨迹数据集上持续优于随机基线并实现高效训练。
正文 · 原文
Abstract:Language models relying on verbalized Chain-of-Thought (CoT) are constrained to token-by-token generation in the language space, where many tokens serve textual coherence rather than core reasoning. This entangles genuine differences in logical trajectories with superficial variations in verbalization. Moving reasoning into a compact, continuous latent space offers a promising alternative for efficiently representing and exploring diverse reasoning trajectories. However, current latent models remain largely deterministic, capturing only a single reasoning trajectory rather than modeling a distribution over multiple valid alternatives. While stochastic approaches parameterize distributions over latent reasoning steps, they frequently suffer from unstable optimization and limited empirical performance. Furthermore, their reliance on autoregressive latent prediction during training precludes effective parallelization, leading to prohibitively long training times. To overcome these bottlenecks, we propose Probabilistic Latent Reasoning Models (PLRM), a parallelizable variational framework for probabilistic latent reasoning. PLRM adopts a JEPA-style objective in which the posterior infers latent distributions in parallel from observed reasoning trajectories, while the prior learns to predict the subsequent latent distribution via teacher forcing. To model heterogeneous reasoning processes, we introduce a discrete strategy variable together with a mixture-based prior, allowing the model to preserve the Gaussian-mixture structure induced by multiple distinct reasoning trajectories rather than collapsing them into a single Gaussian. Across diverse benchmarks, PLRM achieves performance competitive with deterministic latent reasoning models on single-trajectory datasets while enabling efficient training and consistently outperforming competing stochastic baselines on multi-trajectory datasets.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.29385 [cs.LG] |
| (or arXiv:2610.29385v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.29385 arXiv-issued DOI via DataCite |
Submission history
From: Seokhyun Lee [view email]
[v1]
Wed, 21 Oct 2026 08:11:52 UTC (2,039 KB)
[v2]
Thu, 22 Oct 2026 07:41:12 UTC (2,047 KB)
来源：arXiv cs.AI · arxiv.org