zlacker

[parent] [thread] 0 comments
1. Davidz+(OP)[view] [source] 2023-11-20 23:47:59
By the way the next training run of openai will for the first time have a huge amount of data which the model may recognize as being generated by "itself" and therefore will be forced to model itself. Could have implications wrt logic among other things
[go to top]