automated

REAP is an automated pipeline that curates production-derived benchmarks for coding agents from real developer-agent sessions, using LLM-based classification and stability checks to ensure reliable evaluation without manual labeling.

The paper introduces Kernel Discovery, an LLM-driven evolutionary framework for high-dimensional Bayesian optimization that searches a broader kernel space and achieves state-of-the-art results on benchmarks.

This paper introduces a knowledge-based approach using knowledge graph embeddings to automatically assess big data quality by predicting missing edges between context representations and quality rules, outperforming traditional matching methods.