Evaluating the Safety and Skill Reasoning of Large Reasoning Models Under Compute Constraints

Adarsha Balaji, Le Chen, Rajeev Thakur, Franck Cappello, Sandeep Madireddy

Keywords: Large Language Models, Foundation Models, Reasoning, Safety Analysis, Artificial Intelligence

TL;DR: We analyze the skill and safet performance of Large Reasoning Models under compute constraints.

Abstract: Test-time compute scaling has demonstrated the ability to improve the performance of reasoning language models by generating longer chain-of-thought (CoT) sequences. However, this increase in performance comes with an increase in computational cost. In this work, we investigate two compute constraint strategies: (1) reasoning length constraint and (2) model quantization, as methods to reduce the compute demand of reasoning models and study their impact on their safety performance. Specifically, we explore two approaches to apply compute constraints to reasoning models: (1) fine-tuning reasoning models using a length-controlled policy optimization (LCPO) based reinforcement learning method to satisfy a user-defined CoT reasoning length, and (2) applying quantization to maximize the generation of CoT sequences within a user-defined compute constraint. Furthermore, we study the trade-off between the computational efficiency and the safety of the model. We demonstrate that under a fixed compute budget, quantized reasoning models, that reason for longer (more reasoning tokens), perform at par with full-precision reasoning models.

Submission Number: 157