Scaling Model Size for Zero-Shot Cross-Lingual Retrieval on Artificially Code-Switched Data in Low-Resource Settings

Transferring information retrieval (IR) models from a high-resource language (typically English) to other languages in a zero-shot fashion has become a widely adopted approach. In this work, we show that the effectiveness of zero-shot rankers diminishes when queries and documents are present in different languages. Motivated by this, we propose to train ranking models on artificially code-switched data instead, which we generate by utilizing bilingual lexicons. To this end, we experiment with lexicons induced from (1) cross-lingual word embeddings and (2) parallel Wikipedia page titles. We use

Research goal: How does the scaling of model size (e.g., 3B to 175B parameters) influence the effectiveness of zero-shot cross-lingual retrieval when trained on artificially code-switched data, measured by XNLI accuracy and retrieval metrics (e.g., MAP, MRR) for low-resource language pairs?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.6/10.