Latency Versus π0.5 Rollout Performance

A Python pipe buffer was quietly adding hundreds of milliseconds to our π0.5 rollouts. Finding it, then cutting inference from over 230 ms to 87 ms, changed how the robot behaved. Recall that in our

A Python pipe buffer was quietly adding hundreds of milliseconds to our π0.5 rollouts. Finding it, then cutting inference from over 230 ms to 87 ms, changed how the robot behaved. Recall that in our last post, we had ablated various training and rollout parameters to maximize performance. We were after a particular use case – we had a pristine dataset collected on a YAM data collection rig and we wanted to roll a model trained on that rig onto another slightly different, but still YAM-based setu