Product Evaluation

GitHub's secret scanning LLM cuts false positives by 30% with product-focused evaluation

GitHub's LLM for secret scanning reduced false positives by 30% by prioritizing product decisions over model tweaks, according to their evaluation strategy.