Ask HN: Minified version of GPT-2 model?
So, I have been trying to tinker with GPT-2 model on an arbitrary text. The model just seems to be pretty heavy for my machine. Any idea on how we can have minified version of this? I wonder if there are any releases for the model with simplified architecture? (I guess that doesn't make it GPT-2 either if it is minified...). I am trying to use 117M model.
0 comments
[ 3.3 ms ] story [ 6.3 ms ] threadNo comments yet.