ML//model//GPT//GPT-2

GPT-2 is the 1.5B-parameter GPT that OpenAI called too dangerous to release, and whose release it staged in waves. It had no fine-tuning, only pure pre-training, and it could write coherent paragraphs.

GPT-2 is the 1.5B-parameter GPT that OpenAI called too dangerous to release, and whose release it staged in waves. It had no fine-tuning, only pure pre-training, and it could write coherent paragraphs.

In retrospect it looks quaint, but at the time the idea that a model could generate believable text was alarming to some people.