Improving Zero-Shot Cross-Lingual Transfer via Progressive Code-Switching
Li, Zhuoran, Hu, Chunming, Chen, Junfan, Chen, Zhijun, Guo, Xiaohui, Zhang, Richong
–arXiv.org Artificial Intelligence
Code-switching is a data augmentation scheme mixing words from multiple languages into source lingual text. It has achieved considerable generalization performance of cross-lingual transfer tasks by aligning cross-lingual contextual word representations. However, uncontrolled and over-replaced code-switching would augment dirty samples to model training. In other words, the excessive code-switching text samples will negatively hurt the models' cross-lingual transferability. To this end, we propose a Progressive Code-Switching (PCS) method to gradually generate moderately difficult code-switching examples for the model to discriminate from easy to hard. The idea is to incorporate progressively the preceding learned multilingual knowledge using easier code-switching data to guide model optimization on succeeding harder code-switching data. Specifically, we first design a difficulty measurer to measure the impact of replacing each word in a sentence based on the word relevance score. Then a code-switcher generates the code-switching data of increasing difficulty via a controllable temperature variable. In addition, a training scheduler decides when to sample harder code-switching data for model training. Experiments show our model achieves state-of-the-art results on three different zero-shot cross-lingual transfer tasks across ten languages.
arXiv.org Artificial Intelligence
Jun-19-2024
- Country:
- Europe > France (0.04)
- South America > Paraguay
- North America > United States
- New York > New York County > New York City (0.04)
- Asia
- Singapore (0.04)
- Indonesia > Bali (0.04)
- Middle East > UAE
- Abu Dhabi Emirate > Abu Dhabi (0.04)
- India
- Haryana (0.05)
- Uttar Pradesh (0.05)
- China
- Beijing > Beijing (0.04)
- Zhejiang Province > Hangzhou (0.04)
- Shanxi Province (0.04)
- Genre:
- Research Report (0.82)
- Technology: