Meet Phi-1.5, the new language model that could make training AI radically cheaper and faster
By generating curated, high quality, synthetic data using existing LLMs and training a new model on this, the researchers are able to achieve
Source: venturebeat.com