deep learning

GPT-2 Twitter bot

aitextgen I’ve been loosely following developments in machine learning since the AlexNet days back in 2012. I was particularly excited by OpenAI’s release of GPT-2 in 2019 because it seemed like “the solution for everything”. But sadly, even with the help of the code-first deep learning course at fast.ai, using GPT-2 was beyond me. It required ops knowledge I did not have to configure AWS instances and CS knowledge I did not have to tune countless parameters.