Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2402.07610 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909232103686144 |
|---|---|
| author | Wang, Haoyu Ma, Guozheng Meng, Ziqiao Qin, Zeyu Shen, Li Zhang, Zhong Wu, Bingzhe Liu, Liu Bian, Yatao Xu, Tingyang Wang, Xueqian Zhao, Peilin |
| author_facet | Wang, Haoyu Ma, Guozheng Meng, Ziqiao Qin, Zeyu Shen, Li Zhang, Zhong Wu, Bingzhe Liu, Liu Bian, Yatao Xu, Tingyang Wang, Xueqian Zhao, Peilin |
| contents | Self-alignment is an effective way to reduce the cost of human annotation while ensuring promising model capability. However, most current methods complete the data collection and training steps in a single round, which may overlook the continuously improving ability of self-aligned models. This gives rise to a key query: What if we do multi-time bootstrapping self-alignment? Does this strategy enhance model performance or lead to rapid degradation? In this paper, our pioneering exploration delves into the impact of bootstrapping self-alignment on large language models. Our findings reveal that bootstrapping self-alignment markedly surpasses the single-round approach, by guaranteeing data diversity from in-context learning. To further exploit the capabilities of bootstrapping, we investigate and adjust the training order of data, which yields improved performance of the model. Drawing on these findings, we propose Step-On-Feet Tuning (SOFT) which leverages model's continuously enhanced few-shot ability to boost zero or one-shot performance. Based on easy-to-hard training recipe, we propose SOFT+ which further boost self-alignment's performance. Our experiments demonstrate the efficiency of SOFT (SOFT+) across various classification and generation tasks, highlighting the potential of bootstrapping self-alignment on continually enhancing model alignment performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_07610 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Step-On-Feet Tuning: Scaling Self-Alignment of LLMs via Bootstrapping Wang, Haoyu Ma, Guozheng Meng, Ziqiao Qin, Zeyu Shen, Li Zhang, Zhong Wu, Bingzhe Liu, Liu Bian, Yatao Xu, Tingyang Wang, Xueqian Zhao, Peilin Computation and Language Artificial Intelligence Self-alignment is an effective way to reduce the cost of human annotation while ensuring promising model capability. However, most current methods complete the data collection and training steps in a single round, which may overlook the continuously improving ability of self-aligned models. This gives rise to a key query: What if we do multi-time bootstrapping self-alignment? Does this strategy enhance model performance or lead to rapid degradation? In this paper, our pioneering exploration delves into the impact of bootstrapping self-alignment on large language models. Our findings reveal that bootstrapping self-alignment markedly surpasses the single-round approach, by guaranteeing data diversity from in-context learning. To further exploit the capabilities of bootstrapping, we investigate and adjust the training order of data, which yields improved performance of the model. Drawing on these findings, we propose Step-On-Feet Tuning (SOFT) which leverages model's continuously enhanced few-shot ability to boost zero or one-shot performance. Based on easy-to-hard training recipe, we propose SOFT+ which further boost self-alignment's performance. Our experiments demonstrate the efficiency of SOFT (SOFT+) across various classification and generation tasks, highlighting the potential of bootstrapping self-alignment on continually enhancing model alignment performance. |
| title | Step-On-Feet Tuning: Scaling Self-Alignment of LLMs via Bootstrapping |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2402.07610 |