This report synthesises findings from 11 peer-reviewed papers addressing the following research question: What is the degradation rate in HumanEval pass@1 scores for code generation tasks when applying dynamic quantization to transformer attention layers. 13 claims were extracted from source literature; 13 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 degradation rate in HumanEval pass@1 scores for code generation tasks when applying dynamic quantization to transformer attention layers?Autonomous literature synthesis. Automated review score: 8.7/10. Full text and citation available at Assignee Research.