Comment by az226

Mature training pipelines, plus ever expanding RL datasets of increased quality, and mega GPU clusters to finish training in a few weeks. Automated safety and reliability testing.