Cross-lingual Transfer Performance in XTREME-R: Intermediate Task Selection and Model Training Strategies

In zero-shot cross-lingual transfer, a supervised NLP task trained on a corpus in one language is directly applicable to another language without any additional training. A source of cross-lingual transfer can be as straightforward as lexical overlap between languages (e.g., use of the same scripts, shared subwords) that naturally forces text embeddings to occupy a similar representation space. Recently introduced cross-lingual language model (XLM) pretraining brings out neural parameter sharing in Transformer-style networks as the most important factor for the transfer. In this paper, we aim

Research goal: Does the choice of intermediate task (e.g., natural language inference, sentiment analysis) affect the zero-shot cross-lingual transfer performance on the XTREME-R benchmark for typologically diverse languages when comparing English-trained versus multilingual-trained models?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.9/10.