Jonas noted the improvements from FSD (Supervised) V12.3.4 to V12.3.6 are evident, with a more substantial enhancement expected in V12.4, where most training compute limitations are removed.
Describing a situation where cars learn by learning instead of labeling, Jonas emphasized how the FSD system demonstrates heightened sensitivity and acts cautiously or assertively as needed.
Jonas illustrated an example where a robot carries out a complex task based on natural language understanding and observation, highlighting the system's capability to comprehend instructions through previous learning.
The analyst estimated Tesla's data collection from its fleet at about 5.5 million, emphasizing the vast amount of information the company is leveraging to enhance its driving technology.
[
add
]
[
|
|
...
]