GPT3: Unsupervised pre-training + a few* examples is all you need. *Up to 5 (Conversational QA) - 50 examples (Winogrande, PhysicalQA, TriviaQA) https://arxiv.org/abs/2005.14165

Scaling laws: Larger models trained on lesser data* are what you you need. *10x more compute should be spent on 5.5x larger model and 1.8x more tokens https://arxiv.org/abs/2001.08361