GPT-3 / GPT base

OpenAI · text · generative · anchor

Closed-weight autoregressive language model, cited in the corpus in its instruction-tuned text-davinci-003 release.

Note
no anchor recorded: the paper behind it is Brown et al., "Language Models are Few-Shot Learners", NeurIPS 2020, but no search returned its identifier from a source, so writing one here would be memory rather than evidence
Variants
text-davinci-003, Babbage, text-babbage-001, text-davinci-001, text-davinci-002

Findings

Shared mechanisms