Interpreting GPT: the logit lens – LessWrong(2020)

2 points| nalinidash | 4 months ago |lesswrong.com