Hi,
I was just wondering if anyone had attempted to implement the pytorch version of mini gpt2, using the fastai data loaders, or training loop, to try and retrain it on another corpus similar too this.
(Typing this on my phone…apologies if it makes no sense)