Training a Distributed Managed Video Cache

The occupancy network is trained with large, ultra-labeled data set without any human in the loop. All these videos are stored in 30 petabytes of a distributed managed video cache. It takes 1.4 billion frames to train that network,. And if you have 100,000 GPU, it would take one hour. We use 10,000 GPUs for training and run 4,000GPUs for auto labeling. The goal here is to accurately represent the world in 3D space for driving.

Transcript

Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.