RL-STaR: Theoretical Analysis of Reinforcement Learning Frameworks for Self-Taught Reasoner
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917981664051200 |
|---|---|
| author | Chang, Fu-Chieh Lee, Yu-Ting Shih, Hui-Ying Tseng, Yi Hsuan Wu, Pei-Yuan |
| author_facet | Chang, Fu-Chieh Lee, Yu-Ting Shih, Hui-Ying Tseng, Yi Hsuan Wu, Pei-Yuan |
| contents | The reasoning abilities of large language models (LLMs) have improved with chain-of-thought (CoT) prompting, allowing models to solve complex tasks stepwise. However, training CoT capabilities requires detailed reasoning data, which is often scarce. The self-taught reasoner (STaR) framework addresses this by using reinforcement learning to automatically generate reasoning steps, reducing reliance on human-labeled data. Although STaR and its variants have demonstrated empirical success, a theoretical foundation explaining these improvements is lacking. This work provides a theoretical framework for understanding the effectiveness of reinforcement learning on CoT reasoning and STaR. Our contributions are: (1) criteria for the quality of pre-trained models necessary to initiate effective reasoning improvement; (2) an analysis of policy improvement, showing why LLM reasoning improves iteratively with STaR; (3) conditions for convergence to an optimal reasoning policy; and (4) an examination of STaR's robustness, explaining how it can improve reasoning even when incorporating occasional incorrect steps; This framework aims to bridge empirical findings with theoretical insights, advancing reinforcement learning approaches for reasoning in LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_23912 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | RL-STaR: Theoretical Analysis of Reinforcement Learning Frameworks for Self-Taught Reasoner Chang, Fu-Chieh Lee, Yu-Ting Shih, Hui-Ying Tseng, Yi Hsuan Wu, Pei-Yuan Artificial Intelligence Machine Learning The reasoning abilities of large language models (LLMs) have improved with chain-of-thought (CoT) prompting, allowing models to solve complex tasks stepwise. However, training CoT capabilities requires detailed reasoning data, which is often scarce. The self-taught reasoner (STaR) framework addresses this by using reinforcement learning to automatically generate reasoning steps, reducing reliance on human-labeled data. Although STaR and its variants have demonstrated empirical success, a theoretical foundation explaining these improvements is lacking. This work provides a theoretical framework for understanding the effectiveness of reinforcement learning on CoT reasoning and STaR. Our contributions are: (1) criteria for the quality of pre-trained models necessary to initiate effective reasoning improvement; (2) an analysis of policy improvement, showing why LLM reasoning improves iteratively with STaR; (3) conditions for convergence to an optimal reasoning policy; and (4) an examination of STaR's robustness, explaining how it can improve reasoning even when incorporating occasional incorrect steps; This framework aims to bridge empirical findings with theoretical insights, advancing reinforcement learning approaches for reasoning in LLMs. |
| title | RL-STaR: Theoretical Analysis of Reinforcement Learning Frameworks for Self-Taught Reasoner |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2410.23912 |