00:00:00

Share Your Feedback 🏝️

Enhancing Non-Reasoning Models with Reasoning Models

Enhancing Non-Reasoning Models with Reasoning Models

MinWoo(Daniel) Park | Tech Blog

Read more
Previous: Perception LM Next: Reasoning Models Without Thinking

Enhancing Non-Reasoning Models with Reasoning Models

  • Related Project: Private
  • Category: Paper Review
  • Date: 2025-04-23

Enhancing Non-Reasoning Models with Reasoning Models

  • url: https://arxiv.org/abs/2504.09639
  • pdf: https://arxiv.org/pdf/2504.09639
  • html: https://arxiv.org/html/2504.09639v1
  • abstract: Recent advancements in large language models (LLMs), such as DeepSeek-R1 and OpenAI-o1, have demonstrated the significant effectiveness of test-time scaling, achieving substantial performance gains across various benchmarks. These advanced models utilize deliberate “thinking” steps to systematically enhance answer quality. In this paper, we propose leveraging these high-quality outputs generated by reasoning-intensive models to improve less computationally demanding, non-reasoning models. We explore and compare methodologies for utilizing the answers produced by reasoning models to train and improve non-reasoning models. Through straightforward Supervised Fine-Tuning (SFT) experiments on established benchmarks, we demonstrate consistent improvements across various benchmarks, underscoring the potential of this approach for advancing the ability of models to answer questions directly.
Previous: Perception LM Next: Reasoning Models Without Thinking

post contain ""

    No matching posts found containing ""