How to Generate a Programmable Programmable Network

We created a data set of imputatut examples, which contain the language description and then the code. Then we took this massive language model that we pretrain. We find tunit to be good at this particular setting,. Although honestly, i was sceptic and surprised to see that with such small amount of data, we were able toIndeed steer this model, this transformer model, to now become not very good at but started to be reasonable solving some of the easiest problems. And once you see something promising, that is when you start iterating. You start developing new sort of components, new tools, new new agrithmic insites. The results we got, i think