Codestral and Llama3 Pass@1 Performance on LiveCodeBench Time-Split Evaluation

This report synthesises findings from 2 peer-reviewed papers addressing the following research question: How does the pass@1 performance of Codestral compare to Llama3 on LiveCodeBench's time-split evaluation to measure contamination effects in code generation. 11 claims were extracted from source literature; 10 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: How does the pass@1 performance of Codestral compare to Llama3 on LiveCodeBench's time-split evaluation to measure contamination effects in code generation?

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