Scalability of Hybrid Batch Training in Multilingual Language Models Beyond Bloom-7B for XNLI Retrieval Performance
Description
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: How does the hybrid batch training strategy scale to multilingual language models beyond Bloom-7B (e.g., Bloom-176B) in terms of performance trade-offs between monolingual and cross-lingual retrieval accuracy on the XNLI benchmark?
Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.8/10.