No I dont know either. But we have to use the data we have. Perhaps it will stimulate more data creation, especially on Threads when people wake up to the fact of what is taking place.
This is why databases are crucial. We need to build the data for the LE to train on.
The data is there waiting to be mined, and when the announcement is made, people will surely cooperate by creating more data, like when the algorithm, the training will surely be done in a phase in which everyone will have free access to the features it has to guarantee faster and more effective training, surely later a second stage will be launched in which some features will be made premium.
I am not sure of the phase in. I do know they are following the Elon model where the features are available only to those who paid i.e. Premium. At least this will be the prompt engine. Obviously all will be feeding in but I dont think they will be able to pull data out.
It is going to be used to enhance the appeal of Premium. Keep building features until it is foolish not to get it.
Of course, I only speculate with the idea of how to train it, but perhaps they are already doing it and when the node is operational, either implement the characteristics to premium users and that's it.
Well you know more about that stuff than I do. I have no idea how these LLMs are trained or what the Leo team had to create for LeoAI. I can only hope it follows the same trend as other models that keep improving with time.
The final idea is that, to add more value and features that make the use of Leo more attractive and attract people to spend more time on the platform, the way they train it, they will already know that they are the experts, I I'm just curious hahaha, in the end users don't care about the backend, they just want it to look pretty and work.