Performance of XLM-R Models on Zero-Shot Cross-Lingual Semantic Textual Similarity in XTREME-R for Low-Resource African Languages

Pre-trained multilingual language encoders, such as multilingual BERT and XLM-R, show great potential for zero-shot cross-lingual transfer. However, these multilingual encoders do not precisely align words and phrases across languages. Especially, learning alignments in the multilingual embedding space usually requires sentence-level or word-level parallel corpora, which are expensive to be obtained for low-resource languages. An alternative is to make the multilingual encoders more robust; when fine-tuning the encoder using downstream task, we train the encoder to tolerate noise in the contex

Research goal: How do XLM-R models of different sizes (e.g., 1B vs. 10B parameters) perform on zero-shot cross-lingual transfer for semantic textual similarity in XTREME-R when evaluated on adversarial or out-of-domain test sets for low-resource African languages?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.2/10.