gpt1-released-june-2018

IN premise — entries/2026/06/21/wiki-Generative_pre-trained_transformer.md

Created 2026-06-21T09:50:09+00:00

GPT-1 was introduced by OpenAI on June 11, 2018, trained on BookCorpus using decoder-only transformer with generative pre-training

Summary

This pins down the exact date and technical setup of OpenAI's first GPT model, giving the system a concrete baseline for tracking how the architecture and training approach evolved in later releases. It establishes that the original design was a decoder-only transformer trained to generate text from books, which is a distinct starting point from the bidirectional encoder designs popular in that same period.