Posts

All the articles I've posted.

• SlopCodeBench: Measuring Code Erosion Under Iterative Specification Refinement

  G. Orlanski, D. Roy, A. Yun, C. Shin, A. Gu, A. Ge, D. Adila, A. Albarghouthi, and F. Sala

  View code on GitHub Paper View paper on arXiv | 2025 | Technical Report

  SlopCodeBench evaluates AI coding agents under iterative specification updates. Unlike single-shot benchmarks, SCBench reveals verbosity and structural erosion that make agent-written code unmaintainable over time.

• Reward Models Enable Scalable Code Verification by Trading Accuracy for Throughput

  View code on GitHub Paper View paper on arXiv | 2025 | Preprint

  Outcome Reward Models for code verification allow one to trade accuracy for speed in the generate-then-rank paradigm. This can be further improved through a generate-prune-then-rank approach where a weaker verifier prunes solutions prior to ranking, thus saving work on incorrect tokens. We show that this hybrid approach can be 11.65 times faster than running the whole test suite while only being 8.33% less accurate.