arXiv:2605.25230 · v3
Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models
摘要
Recent work on recursive architectures has shown that tiny neural networks can be surprisingly powerful on structured reasoning tasks. The trick is to model reasoning trajectories with a latent dynamical system. We argue that the inference-time behaviour of these architectures is best understood as approximate inference over latent reasoning trajectories, with deterministic recursion as the one-particle, zero-noise limit. We make this view operational through guided reasoning: stochastic perturbations of the reasoning dynamics propose neighbouring trajectories, and the model's existing early-stopping head reweights them online. The framework yields three diagnostics computable from inference traces: local stability, guide alignment, with a label-free necessary condition, and cloud-token entropy. These predict whether the procedure will help and which of its outputs to trust. On Sudoku-Extreme it lifts exact-solve accuracy from to without retraining, and on PPBench, a four-type pencil-puzzle benchmark, from to with no tuning; on Maze-Hard the diagnostics flag a misaligned guide, as validation performance later confirms. The same machinery thus characterises both when recursive reasoning has room to improve at the trajectory level and when the model's internal guide can recover said improvements. Code is available at `https://github.com/digiLab-ai/bigr`.
作者备注:Accepted at NeurIPS 2026, main track. An earlier version was presented at the ICML 2026 Workshop on Structured Probabilistic Inference and Generative Modeling (SPIGM), Seoul. Code: https://github.com/digiLab-ai/bigr
期刊信息:Advances in Neural Information Processing Systems 40 (NeurIPS 2026)