Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2506.15050 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915348957102080 |
|---|---|
| author | Fan, Tiantian Liu, Lingjun Yue, Yu Chen, Jiaze Wang, Chengyi Yu, Qiying Zhang, Chi Lin, Zhiqi Zhu, Ruofei Yuan, Yufeng Zuo, Xiaochen Ma, Bole Zhang, Mofan Liu, Gaohong Zhang, Ru Zhou, Haotian Xie, Cong Zhu, Ruidong Zhang, Zhi Liu, Xin Wang, Mingxuan Yan, Lin Wu, Yonghui |
| author_facet | Fan, Tiantian Liu, Lingjun Yue, Yu Chen, Jiaze Wang, Chengyi Yu, Qiying Zhang, Chi Lin, Zhiqi Zhu, Ruofei Yuan, Yufeng Zuo, Xiaochen Ma, Bole Zhang, Mofan Liu, Gaohong Zhang, Ru Zhou, Haotian Xie, Cong Zhu, Ruidong Zhang, Zhi Liu, Xin Wang, Mingxuan Yan, Lin Wu, Yonghui |
| contents | Recently, test-time scaling Large Language Models (LLMs) have demonstrated exceptional reasoning capabilities across scientific and professional tasks by generating long chains-of-thought (CoT). As a crucial component for developing these reasoning models, reinforcement learning (RL), exemplified by Proximal Policy Optimization (PPO) and its variants, allows models to learn through trial and error. However, PPO can be time-consuming due to its inherent on-policy nature, which is further exacerbated by increasing response lengths. In this work, we propose Truncated Proximal Policy Optimization (T-PPO), a novel extension to PPO that improves training efficiency by streamlining policy update and length-restricted response generation. T-PPO mitigates the issue of low hardware utilization, an inherent drawback of fully synchronized long-generation procedures, where resources often sit idle during the waiting periods for complete rollouts. Our contributions are two-folds. First, we propose Extended Generalized Advantage Estimation (EGAE) for advantage estimation derived from incomplete responses while maintaining the integrity of policy learning. Second, we devise a computationally optimized mechanism that allows for the independent optimization of the policy and value models. By selectively filtering prompt and truncated tokens, this mechanism reduces redundant computations and accelerates the training process without sacrificing convergence performance. We demonstrate the effectiveness and efficacy of T-PPO on AIME 2024 with a 32B base model. The experimental results show that T-PPO improves the training efficiency of reasoning LLMs by up to 2.5x and outperforms its existing competitors. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_15050 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Truncated Proximal Policy Optimization Fan, Tiantian Liu, Lingjun Yue, Yu Chen, Jiaze Wang, Chengyi Yu, Qiying Zhang, Chi Lin, Zhiqi Zhu, Ruofei Yuan, Yufeng Zuo, Xiaochen Ma, Bole Zhang, Mofan Liu, Gaohong Zhang, Ru Zhou, Haotian Xie, Cong Zhu, Ruidong Zhang, Zhi Liu, Xin Wang, Mingxuan Yan, Lin Wu, Yonghui Artificial Intelligence Recently, test-time scaling Large Language Models (LLMs) have demonstrated exceptional reasoning capabilities across scientific and professional tasks by generating long chains-of-thought (CoT). As a crucial component for developing these reasoning models, reinforcement learning (RL), exemplified by Proximal Policy Optimization (PPO) and its variants, allows models to learn through trial and error. However, PPO can be time-consuming due to its inherent on-policy nature, which is further exacerbated by increasing response lengths. In this work, we propose Truncated Proximal Policy Optimization (T-PPO), a novel extension to PPO that improves training efficiency by streamlining policy update and length-restricted response generation. T-PPO mitigates the issue of low hardware utilization, an inherent drawback of fully synchronized long-generation procedures, where resources often sit idle during the waiting periods for complete rollouts. Our contributions are two-folds. First, we propose Extended Generalized Advantage Estimation (EGAE) for advantage estimation derived from incomplete responses while maintaining the integrity of policy learning. Second, we devise a computationally optimized mechanism that allows for the independent optimization of the policy and value models. By selectively filtering prompt and truncated tokens, this mechanism reduces redundant computations and accelerates the training process without sacrificing convergence performance. We demonstrate the effectiveness and efficacy of T-PPO on AIME 2024 with a 32B base model. The experimental results show that T-PPO improves the training efficiency of reasoning LLMs by up to 2.5x and outperforms its existing competitors. |
| title | Truncated Proximal Policy Optimization |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2506.15050 |