CLLGJul 7

Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization

arXiv:2607.0586118.1
Predicted impact top 35% in CL · last 90 daysOriginality Incremental advance
AI Analysis

For developers of large reasoning models, this work addresses a failure mode where explicit thinking degrades factual accuracy, offering a method to improve factual reliability without sacrificing reasoning performance.

Large reasoning models (LRMs) can suffer from thinking-induced hallucination where explicit thinking overturns correct non-thinking answers. The proposed MARGO framework uses mixed-mode advantage regularization to suppress hallucination-prone thinking, improving factual reliability on QA benchmarks while preserving general reasoning ability.

Large reasoning models (LRMs) improve language model capabilities by generating explicit thinking traces before final answers. In factuality-oriented question answering (QA), such thinking often improves overall performance by helping the model recover relevant knowledge and refine its answers. However, we find that this benefit is not uniform at the instance level: explicit thinking can also overturn correct non-thinking answers and lead to factual drift. We refer to this failure mode as \emph{thinking-induced hallucination}. To explain this phenomenon, we formulate explicit thinking in factuality QA as a thinking residual over the model's direct-answer tendency, which can either recover missing knowledge or introduce unsupported associations. Based on this formulation, we propose MARGO, \underline{\textit{M}}ixed-Mode \underline{\textit{A}}dvantage \underline{\textit{R}}egularization for \underline{\textit{G}}rounded \underline{\textit{O}}ptimization, a reinforcement learning framework that uses non-thinking rollouts as same-model references in advantage estimation. By constructing mixed-mode rollout groups with both thinking and non-thinking trajectories, MARGO evaluates whether explicit thinking adds factual value beyond direct answering, thereby suppressing hallucination-prone thinking while preserving beneficial thinking behaviors. Experiments across multiple factuality-oriented QA benchmarks demonstrate that MARGO improves factual reliability over strong baselines, while evaluations on mathematical benchmarks show that it preserves general reasoning ability.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes