Our experiments are conducted in an unsupervised goal-conditioned setting, where no demonstrations or rewards are provided, so an agent must explore (from scratch) and learn how to maximize the likelihood of reaching commanded goals.

My Highlights

performance consistently improves as scale increases in complex tasks. In addition, deep models exhibit qualitatively better behaviors which might be interpreted as implicitly acquired skills necessary to reach the goal.

AI Model Scaling and Performance

Character counts are represented on low-dimensional curved manifolds discretized by sparse feature families, analogous to biological place cells.

When Models Manipulate Manifolds: The Geometry of a Counting Task

Arena's mission is to measure and advance the frontier of AI through open, rigorous, and community-driven evaluation.

Arena Academic Partnerships: Funding AI Evaluation Research