Inertia-1: Open Exploration of Wearable Motion Foundation Models

Summary

Inertia-1 is a fully open exploration of wearable motion foundation models from the Yang AI Lab (UCLA / Yuzhe Yang + Sriram Sankararaman). Using 18.2M hours of accelerometer data from >115,000 people across 15 datasets, it provides a controlled framework studying data choices, model architectures, and training objectives. Key framing: a comprehensive cookbook for building motion FMs that generalize across tasks and sensing conditions.