SPREG: Structured Plan Repair with Entropy-Guided Test-Time Intervention for Large Language Model Reasoning
This work addresses the problem of logical hallucinations and drift in long-chain reasoning for LLMs, offering a practical test-time intervention that improves accuracy without retraining.
SPREG introduces a lightweight inference-time framework that uses entropy monitoring and dynamic repair to correct logical errors in LLM reasoning, achieving a 20.0% absolute accuracy improvement on AIME25.
Large Language Models (LLMs) are prone to logical hallucinations and stochastic drifts during long-chain reasoning. While Classifier-Free Guidance (CFG) can improve instruction adherence, standard static implementations often cause semantic dilution and linguistic degradation. We propose SPREG (Structured Plan-guided Real-time Entropy Gating), a lightweight inference-time framework for surgical error rectification. SPREG employs an adaptive dual-threshold mechanism to monitor real-time entropy, identifying sudden ``entropy spikes'' as reliable indicators of logical failure. Upon detection, it triggers a dynamic repair by replacing uninformative null-priors with reference distributions synthesized from historical high-confidence states. By modulating guidance intensity according to structured reasoning stages (e.g., Action, Observation), SPREG steers the model back to a stable manifold without compromising fluency. Our experiments demonstrate significant gains, notably a 20.0% absolute accuracy improvement on AIME25, while effectively suppressing uncontrolled entropy drift in complex tasks.