Case Study: Woven by Toyota accelerates autonomous driving learning with Weights & Biases

Woven by Toyota boosts ML velocity 10x with Weights & Biases

Woven by Toyota, a mobility technology subsidiary of Toyota, faced significant challenges in scaling its autonomous driving development. Their original workflow for data collection, manual curation, and annotation was slow and inefficient, taking over a year to complete a cycle. They needed to drastically accelerate this process to achieve continuous learning and ensure safety, requiring a 10x improvement in velocity.

By implementing Weights & Biases for experiment tracking, sweeps, and reporting, the team automated tedious processes and gained critical tractability. This new system, part of a larger overhaul that included leveraging customer vehicle data and autolabeling, reduced their model development cycle from 20 months to just 2 months. Weights & Biases provided a central system of record that enabled faster collaboration and a 10x increase in team velocity, with tools like Tables being used to create a cross-functional model leaderboard.

Woven by Toyota