few-shot
adj1
A machine learning paradigm where a model is trained on a very small amount of data, typically less than that required for traditional machine learning approaches.
Machine learning
- No additional prompts for getting the responses were used, i.e. the data was created with a basic prompt in a zero-shot scenario. This is in contrast to the benchmarks by OpenAI, who used an engineered prompt in a few-shot scenario to guide the generation of essays.2023 October 30, Herbold et al., “A large-scale comparison of human-written versus ChatGPT-generated essays”, in Scientific Reports, volume 13, page 4: