Knowledge Commons of Institute of Automation,CAS
TBERT: Dynamic BERT Inference with Top-k Based Predictors | |
Liu, Zejian1,2; Zhao, Kun1; Cheng, Jian1,2,3 | |
2023-04 | |
会议名称 | Design, Automation & Test in Europe Conference |
会议日期 | 2023-4-17 |
会议地点 | Antwerp, Belgium |
摘要 | Dynamic inference is a compression method that adaptively prunes unimportant components according to the input at the inference stage, which can achieve a better tradeoff between computational complexity and model accuracy than static compression methods. However, there are two limitations in previous works. The first one is that they usually need to search the threshold on the evaluation dataset to achieve the target compression ratio, but the search process is non-trivial. The second one is that these methods are unstable. Their performance will be significantly degraded on some datasets, especially when the compression ratio is high. In this paper, we propose TBERT, a simple yet stable dynamic inference method. TBERT utilizes the top-k-based pruning strategy which allows accurate control of the compression ratio. To enable stable end-to-end training of the model, we carefully design the structure of the predictor. Moreover, we propose adding auxiliary classifiers to help the model’s training. Experimental results on the GLUE benchmark demonstrate that our method achieves higher performance than previous state-of-the-art methods. |
关键词 | Transformer Dynamic Inference Pruning |
学科门类 | 工学::计算机科学与技术(可授工学、理学学位) |
收录类别 | EI |
语种 | 英语 |
七大方向——子方向分类 | AI芯片与智能计算 |
国重实验室规划方向分类 | 其他 |
是否有论文关联数据集需要存交 | 否 |
文献类型 | 会议论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/52036 |
专题 | 复杂系统认知与决策实验室_高效智能计算与学习 |
通讯作者 | Cheng, Jian |
作者单位 | 1.National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences 2.School of Future Technology, University of Chinese Academy of Sciences 3.AiRiA |
第一作者单位 | 模式识别国家重点实验室 |
通讯作者单位 | 模式识别国家重点实验室 |
推荐引用方式 GB/T 7714 | Liu, Zejian,Zhao, Kun,Cheng, Jian. TBERT: Dynamic BERT Inference with Top-k Based Predictors[C],2023. |
条目包含的文件 | 下载所有文件 | |||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
TBERT_ Dynamic BERT (3426KB) | 会议论文 | 开放获取 | CC BY-NC-SA | 浏览 下载 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论