175B-parameter autoregressive language model demonstrating strong few-shot learning.
OpenAI published GPT-3, demonstrating few-shot learning at 175 billion parameters.
