Skywork Open Reasoner 1 Technical Report
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913865817653248 |
|---|---|
| author | He, Jujie Liu, Jiacai Liu, Chris Yuhao Yan, Rui Wang, Chaojie Cheng, Peng Zhang, Xiaoyu Zhang, Fuxiang Xu, Jiacheng Shen, Wei Li, Siyuan Zeng, Liang Wei, Tianwen Cheng, Cheng An, Bo Liu, Yang Zhou, Yahui |
| author_facet | He, Jujie Liu, Jiacai Liu, Chris Yuhao Yan, Rui Wang, Chaojie Cheng, Peng Zhang, Xiaoyu Zhang, Fuxiang Xu, Jiacheng Shen, Wei Li, Siyuan Zeng, Liang Wei, Tianwen Cheng, Cheng An, Bo Liu, Yang Zhou, Yahui |
| contents | The success of DeepSeek-R1 underscores the significant role of reinforcement learning (RL) in enhancing the reasoning capabilities of large language models (LLMs). In this work, we present Skywork-OR1, an effective and scalable RL implementation for long Chain-of-Thought (CoT) models. Building on the DeepSeek-R1-Distill model series, our RL approach achieves notable performance gains, increasing average accuracy across AIME24, AIME25, and LiveCodeBench from 57.8% to 72.8% (+15.0%) for the 32B model and from 43.6% to 57.5% (+13.9%) for the 7B model. Our Skywork-OR1-32B model surpasses both DeepSeek-R1 and Qwen3-32B on the AIME24 and AIME25 benchmarks, while achieving comparable results on LiveCodeBench. The Skywork-OR1-7B and Skywork-OR1-Math-7B models demonstrate competitive reasoning capabilities among models of similar size. We perform comprehensive ablation studies on the core components of our training pipeline to validate their effectiveness. Additionally, we thoroughly investigate the phenomenon of entropy collapse, identify key factors affecting entropy dynamics, and demonstrate that mitigating premature entropy collapse is critical for improved test performance. To support community research, we fully open-source our model weights, training code, and training datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_22312 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Skywork Open Reasoner 1 Technical Report He, Jujie Liu, Jiacai Liu, Chris Yuhao Yan, Rui Wang, Chaojie Cheng, Peng Zhang, Xiaoyu Zhang, Fuxiang Xu, Jiacheng Shen, Wei Li, Siyuan Zeng, Liang Wei, Tianwen Cheng, Cheng An, Bo Liu, Yang Zhou, Yahui Machine Learning Artificial Intelligence Computation and Language The success of DeepSeek-R1 underscores the significant role of reinforcement learning (RL) in enhancing the reasoning capabilities of large language models (LLMs). In this work, we present Skywork-OR1, an effective and scalable RL implementation for long Chain-of-Thought (CoT) models. Building on the DeepSeek-R1-Distill model series, our RL approach achieves notable performance gains, increasing average accuracy across AIME24, AIME25, and LiveCodeBench from 57.8% to 72.8% (+15.0%) for the 32B model and from 43.6% to 57.5% (+13.9%) for the 7B model. Our Skywork-OR1-32B model surpasses both DeepSeek-R1 and Qwen3-32B on the AIME24 and AIME25 benchmarks, while achieving comparable results on LiveCodeBench. The Skywork-OR1-7B and Skywork-OR1-Math-7B models demonstrate competitive reasoning capabilities among models of similar size. We perform comprehensive ablation studies on the core components of our training pipeline to validate their effectiveness. Additionally, we thoroughly investigate the phenomenon of entropy collapse, identify key factors affecting entropy dynamics, and demonstrate that mitigating premature entropy collapse is critical for improved test performance. To support community research, we fully open-source our model weights, training code, and training datasets. |
| title | Skywork Open Reasoner 1 Technical Report |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2505.22312 |