TL;DR Self-training systems typically degenerate because they lack an external criterion for judging data quality, which opens the door to reward hacking and semantic drift. The paper presents a proof-of-concept …
TL;DR Models trained on blocks from the region of an Ulam spiral around 500m outperform models trained on blocks below 25m, implying there is more easily learnable order in the former region than the latter. Precision and recall scores diverge by region: the …
TL;DR This paper proposes a framework in which AI agents generate and validate new knowledge directly through interaction with their environment, rather than through further human-curated data, resulting in recursive …
