# Mitigating Shortcut Reasoning in Language Models: A Gradient-Aware Training Approach Source: [Mitigating Shortcut Reasoning in Language Models: A Gradient-Aware Training Approach]() Collection index: [llms.txt]() Data updated: 2026-03-25T05:00:04.214Z Authors: Hongyu Cao, Kunpeng Liu, Dongjie Wang, Yanjie Fu Published: 2026-03-21T18:17:17.000Z ## Abstract Large language models exhibit strong reasoning capabilities, yet often rely on shortcuts such as surface pattern matching and answer memorization rather than genuine logical inference. We propose Shortcut-Aware Reasoning Training (SART), a gradient-aware framework that detects and mitigates shortcut-promoting samples via ShortcutScore and gradient surgery. Our method identifies shortcut signals through gradient misalignment with validation objectives and answer-token concentration, and modifies training dynamics accordingly. Experiments on controlled reasoning benchmarks show that SART achieves +16.5% accuracy and +40.2% robustness over the strongest baseline, significantly improving generalization under distribution shifts. Code is available at: https://github.com/fuyanjie/short-cut-aware-data-centric-reasoning. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Reasoning and Shortcut Detection on Three synthetic benchmarks Average (test)](): Reasoning and Shortcut Detection; Three synthetic benchmarks Average (test); Accuracy: 92.5. ## Original sources and implementation [Original paper]() [PDF]()