momo3605
Active Member
Ok...I’m going to ask a really stupid question but... I understand how billions of miles of learning can fit on Elon’s mainframe back at HQ....but how does it all fit on the laptop in your car?
If you’re serious, the billions of miles of labeled photos/videos is used to train models on big GPU racks. Once the model is trained, you deploy the much smaller model on the car’s FSD chip for inference.
Here’s additional info on the difference between training and inference. What’s the Difference Between Deep Learning Training and Inference? | NVIDIA Blog