Scaling Model Size and Self-Invoking Code Generation in Cross-Lingual Tasks

This report synthesises findings from 13 peer-reviewed papers addressing the following research question: What is the impact of scaling model size on self-invoking code generation performance in cross-lingual tasks, as evaluated by pass@1 on MBPP+ when comparing models with <10B, 10B-30B, and >30B. 9 claims were extracted from source literature; 9 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.7/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: What is the impact of scaling model size on self-invoking code generation performance in cross-lingual tasks, as evaluated by pass@1 on MBPP+ when comparing models with <10B, 10B-30B, and >30B parameters?

Autonomous literature synthesis. Automated review score: 8.7/10. Full text and citation available at Assignee Research.