Intel / article

QwQ-32B: Embracing the Power of Reinforcement Learning

Source: qwenlm.github.io
Date: Jul 21, 2026

Why it matters

How scaling RL boosts reasoning in large language models
Context on DeepSeek R1's cold-start data and multi-stage training approach
Qwen's research angle on RL scalability for model intelligence

Key quotes

“Scaling Reinforcement Learning (RL) has the potential to enhance model performance beyond conventional pretraining and post-training methods.”

“For instance, DeepSeek R1 has achieved state-of-the-art performance by integrating cold-start data and multi-stage training, enabling deep thinking and complex reasoning.”

Transcript

QWEN CHAT Hugging Face ModelScope DEMO DISCORD Scaling Reinforcement Learning (RL) has the potential to enhance model performance beyond conventional pretraining and post-training methods. Recent studies have demonstrated that RL can significantly improve the reasoning capabilities of models. For instance, DeepSeek R1 has achieved state-of-the-art performance by integrating cold-start data and multi-stage training, enabling deep thinking and complex reasoning. Our research explores the scalability of Reinforcement Learning (RL) and its impact on enhancing the intelligence of large language models.

Read the source ↗qwenlm.github.io