In-context Learning

In-context learning is the ability of a model to adapt its output based on examples or instructions provided in the prompt, without updating the model’s parameters. It is closely associated with few-shot prompting, where the prompt demonstrates the desired task or format.