GBA: A Tuning-free Approach to Switch between Synchronous and Asynchronous Training for Recommendation Models

Open in new window