Tesla FSD Beta Gets Major Training Boost as Dojo Supercomputer Turns On

Image: Tesla

The Tesla Dojo supercomputer has been launched. Now FSD Beta will receive a significant boost in training. Dojo will increase the speed and accuracy of training by ensuring that autonomous driving is achieved as quickly as possible.

According to Tim Zaman, AI Infra at Tesla, Tesla turned on its Dojo supercomputer on Monday. Its goal is to increase the speed and accuracy of training. Dojo is a supercomputer being developed by Tesla's team to increase the training speed of neural networks on the server side.

“Tesla AI 10k H100 cluster, go live Monday. Due to real-world video training, we may have the largest training datasets in the world, hot tier cache capacity beyond 200PB - orders of magnitudes more than LLMs.
Join us!” wrote Zaman.

Tesla AI 10k H100 cluster, go live monday.
Due to real-world video training, we may have the largest training datasets in the world, hot tier cache capacity beyond 200PB - orders of magnitudes more than LLMs.
Join us!https://t.co/F4A0Qb0CXG

— Tim Zaman (@tim_zaman) August 26, 2023

In late June, Tesla announced it had completed work on its Dojo supercomputer and would put it into production in July. This meant the company was just one step away from powerful hardware that would bring full autonomy to Tesla vehicles. With the launch of the supercomputer developed in-house, many obstacles were overcome that separated the company from fully autonomous driving.

In June, Elon Musk also stated that Dojo had already “been online & running useful tasks for a few months.” However, it remains unknown exactly what he meant. Perhaps at that time, it was only partially launched. However, it is expected that now Dojo has been launched at full capacity.

During the Q2 2023 Earnings Call, Elon Musk talked about the development of Full Self-Driving (FSD) Beta. He said that to achieve level 5 autonomous driving, high computing power is needed. It would provide faster, rapid training of neural networks (NN). Until recently, this has been a fundamental limitation of progress.

“The fundamental rate limiter on the progress of full self-driving is training,” said Musk. “If we had more training compute, we could get it done faster.”

Now, with the launch of Dojo, this problem has been solved and the learning rate of FSD will be much faster. Now the company is working on FSD based on artificial intelligence.

© 2023, Eva Fox | Tesmanian. All rights reserved.

_____________________________

We appreciate your readership! Please share your thoughts in the comment section below. 

Article edited by @SmokeyShorts; follow him on Twitter

Follow @EvaFoxU