In robotics, given all the heterogeneity in the data, if you want to pretrain in a similar manner, we need a different architecture.
Our dream is to have a universal robot brain that you could download and use for your robot without any training at all.
Instead of focused data, we looked to models like GPT-4 for a brute force data approach to problem solving.
The larger the transformer, the better the output, which is critical in improving the adaptability of robots to various challenges.
Collection
[
|
...
]