#in-context-learning

Question 1: Single choice

mediumAI-checkedLLM Fundamentals & Prompting

A pretrained language model answers a task much better after you add a few worked examples (few-shot prompting) or ask it to reason step by step (chain-of-thought prompting). What has changed?

The examples in the prompt trigger a small gradient update of the model's weights1
The model stores the examples and uses them for every later user2
The model is fine-tuned on the prompt before it generates the answer3
Only the input the model conditions on; its weights are the same as before4

#in-context-learning

Timed test