When primed with only a handful of training samples, very large, pretrained language models such as GPT-3 have shown competitive results when compared to fully-supervised, fine-tuned, large, pretrained language models.
Fantastically ordered prompts and where to find them: Overcoming few-shot prompt order sensitivity
![](https://mdrresearch.co.uk/wp-content/uploads/2024/03/drew-dizzy-graham-14lr-KjoyDU-unsplash-1024x683.jpg)