Tesla is preparing to update its Full Self-Driving (FSD) system to improve its autonomous driving capabilities. With the twelfth version of FSD, a much more advanced learning network will be deployed. It was even stated that Tesla will launch the 300 million dollar system with 10 thousand Nvidia graphics cards.
Tesla FSD 12 powered by 10k Nvidia H100 GPU
Tesla, one of the leading companies in the world of autonomous driving, is preparing to build a new system for real-time autonomous training. The company, which has purchased 10 thousand Nvidia H100s worth 30 thousand dollars in recent months, wants to integrate these graphics cards into autonomy with FSD 12.
Making a statement on the subject, Tesla engineer Tim Zaman stated that “they will have the largest data set in the world”. He tweeted, “With real-time video training, we could have the largest training datasets in the world, over 200PB of cache capacity. Much larger levels than LLM,” he said.
LLM, which the Tesla engineer refers to, stands for Large Language Model. It refers to artificial intelligence models that are usually trained using large text data. These models are used in natural language processing and machine learning.
Tesla FSD’s data capacity of 200 petabytes exceeds the training of LLMs. This shows that Tesla has a much more advanced data set compared to models such as ChatGPT and Bard.
Musk has previously stated that they will introduce full autonomy with version 12. This means that self-driving cars will not need human supervision. Let us remind you that Musk recently took a test drive with FSD 12 on live broadcast.
Don’t forget to share your views with us in the comments section…
{{user}} {{datetime}}
{{text}}