SimCSE Performance in Low-Resource Languages vs. Multilingual InfoNCE Methods

This report synthesises findings from 14 peer-reviewed papers addressing the following research question: To what extent does the SimCSE framework maintain Spearman correlation performance on the STS-Benchmark when applied to low-resource languages compared to multilingual adaptations of InfoNCE-based. 11 claims were extracted from source literature; 11 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 9.2/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: To what extent does the SimCSE framework maintain Spearman correlation performance on the STS-Benchmark when applied to low-resource languages compared to multilingual adaptations of InfoNCE-based methods?

Autonomous literature synthesis. Automated review score: 9.2/10. Full text and citation available at Assignee Research.