Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Richmond Sin Jing Xuan, Rishabh Bhardwaj, Soujanya Poria

TL;DR
Post-Reasoning enhances instruction-tuned models by conditioning them to justify answers after response generation, improving performance without additional latency or token cost across diverse benchmarks.
Contribution
It introduces Post-Reasoning, a method that improves model performance by post-hoc justification conditioning, and demonstrates its effectiveness through extensive evaluation and supervised tuning.
Findings
Post-Reasoning improves performance in over 88% of settings.
Achieves a mean relative improvement of 17.37%.
Supervised post-reason tuning further enhances results, exceeding baseline by 8.01%.
Abstract
As the widespread adoption of Large Language Models (LLMs) accelerates, token consumption from intermediate reasoning traces increasingly contributes to inference latency and operational cost. Recent studies suggest that many real-world tasks require little to no explicit reasoning, with additional reasoning sometimes even degrading performance. In this work, we propose \textbf{Post-Reasoning}, a simple yet effective approach that improves instruction-tuned models by conditioning them to justify their answers after generating the final response. By design, it enables the final answer to be obtained without additional latency or token cost, while still improving performance through simple instruction augmentation. We evaluate Post-Reasoning across \(117\) model--benchmark settings spanning \(13\) open and proprietary models, \(4\) model families, and \(9\) diverse reasoning and…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
