Closing the loop on post-training.

Introducing Actuator: a closed-loop control layer for supervised fine-tuning, compression, and reinforcement learning.

Plug it right into your existing stack, monitor what you care about, and adapt automatically when training drifts.