Alec Radford and colleagues at OpenAI announced the original GPT, later known as GPT-1, with the paper "Improving Language Understanding by Generative Pre-Training". Using a two-stage approach—unsupervised pre-training of a transformer on large amounts of text via next-word prediction, followed by fine-tuning on small supervised datasets—it achieved state-of-the-art results on diverse language understanding tasks including textual entailment, commonsense reasoning, and reading comprehension. The model and code were released. It was the first model to carry the GPT name.