GPT-3

wikipedia:GPT-3 (Jun 2020)

wikipedia:Generative Pre-trained Transformer 3

The architecture is a decoder-only transformer network with a 2048-token-long context and 175 billion parameters, requiring 800GB to store.

• Deep learning
• NLP

• Transformer (machine learning model)
• ChatGPT
• GPT-2

See also

• GPT, GPT-2, GPT-3, GPT-4, GPT-4o, Tiktoken, Bigram, Transformer, PaLM, ChatGPT
• NLP, GPT-3, Amazon Comprehend
• OpenAI, GitHub Copilot, ChatGPT, OpenAI Codex, GPT-3, GPT-4, Whisper, Sam Altman, Mira Murati, Greg Brockman, Ilya Sutskever, OpenAI board, John Schulman