gpt-3-175b-few-shot-zero-shot
IN premise — entries/2026/06/21/wiki-Generative_pre-trained_transformer-chunk-1.md
Created 2026-06-21T09:50:09+00:00
GPT-3 had 175 billion parameters and was a breakthrough in few-shot and zero-shot learning.
Summary
GPT-3's 175-billion-parameter scale was large enough that the model could figure out how to perform a new task from just a few examples or even none at all, eliminating the need to retrain it separately for each job. This fact is held as an active anchor in the system because it shapes how we reason about what large language models can and cannot do in general.