← Dictionary / Few-shot learning

Few-shot learning

Appears in 2 papers

The ability to perform a task from a small number of examples — typically tens to hundreds.

As used in Paper 10 — Improving Language Understanding by Generative Pre-Training →

The ability to perform a task from a small number of examples — typically tens to hundreds. GPT-1 still required fine-tuning; GPT-3 demonstrated in-context few-shot learning with no weight updates.

As used in Paper 12 — Language Models are Few-Shot Learners →

Performing a task with 2–5 examples in the prompt. The most practical approach for GPT-3, balancing the clarity from multiple examples with the brevity of a short prompt.

Appears in papers

Paper 10 — Improving Language Understanding by Generative Pre-Training → Paper 12 — Language Models are Few-Shot Learners →