What is few-shot prompting?
Why C is the answer
Giving worked examples in the prompt
Few-shot prompts show the task through examples.
In detail
- Few-shot prompting shows the task through examples.
- Choose examples that cover the hard cases.
Use case
An illustrative scenario, not a real incident.
A prompt includes two worked examples, and the model copies their format.
See it in different ways
flowchart TD
A{"What does the model need?"} -->|"Clear task only"| B["Zero-shot"]
A -->|"An example format"| C["Few-shot"]
A -->|"Reasoning"| D["Chain of thought"]stateDiagram-v2 [*] --> Tokenize Tokenize --> Pretraining Pretraining --> FineTuning FineTuning --> RLHF RLHF --> Evaluation Evaluation --> [*]
flowchart LR S0["Example 1"] --> S1["Example 2"] --> S2["New input"] --> S3["Output in the same style"]
Key point: Few-shot prompting teaches by example.
Why the other options are wrong
- A. Limiting the model to a few output tokens at a time
- That is an output limit.
- B. Sending several short prompts in parallel to the model
- That is about concurrency.
- D. Training a new model with only a few epochs of data
- That is training.