JaCoText Inference Overhead Scaling on Adversarially Perturbed Code Benchmarks

This report synthesises findings from 8 peer-reviewed papers addressing the following research question: How does the computational overhead for inference in JaCoText scale with model size when evaluated on adversarially perturbed programming language benchmarks, measured in terms of latency and. 6 claims were extracted from source literature; 6 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 computational overhead for inference in JaCoText scale with model size when evaluated on adversarially perturbed programming language benchmarks, measured in terms of latency and throughput?

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