论文地址:Improved Automatic Summarization of Subroutines via Attention to File Context
添加文件内上下文作为模型的补充,而不是横向对比
- 引言和相关工作较为清晰
- 该实验中Transformer表现较差,原因未解释
- 指出code工作可能不能用预训练方法,有待商榷
论文地址:Improved Automatic Summarization of Subroutines via Attention to File Context