Impact of Batch Size Scaling in Hybrid Training on Zero-Shot Cross-Lingual and Monolingual Retrieval Trade-offs

Information retrieval across different languages is an increasingly important challenge in natural language processing. Recent approaches based on multilingual pre-trained language models have achieved remarkable success, yet they often optimize for either monolingual, cross-lingual, or multilingual retrieval performance at the expense of others. This paper proposes a novel hybrid batch training strategy to simultaneously improve zero-shot retrieval performance across monolingual, cross-lingual, and multilingual settings while mitigating language bias. The approach fine-tunes multilingual lang

Research goal: What is the impact of scaling the batch size in the hybrid training strategy on the trade-off between zero-shot cross-lingual F1 scores (XTREME-R) and monolingual MAP scores (MS MARCO) in multilingual dense retrieval models?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.7/10.