Self-improving AI harnesses are real and operational, with systems like Shanghai AI Lab's Self-Harness (arXiv 2606.09498) boosting held-out pass rates on Terminal-Bench-2 from 40.5% to 61.9% on MiniMa…
Meta AI's Muse Spark model achieved a perfect score of 30/30 on the Asian Physics Olympiad theoretical exam, surpassing the typical gold medal threshold of 21-23 points. The result, if verified, would…
Shanghai AI Lab has developed Agents-A1, a 35-billion-parameter model that matches trillion-parameter models on multi-step agent tasks. Instead of scaling parameter count, the researchers scaled the t…
Scaling laws predict training loss as model size, dataset size, and compute scale, but their practical application is sensitive to hyperparameter choices like learning rate. Lilian Weng's post highlig…