Understanding Hugging Face within the PyTorch context
Hugging Face [1] is a rapidly growing multi-faceted AI company. On the one hand, it provides a host of libraries related to training, evaluating, optimizing, and deploying AI models. On the other hand, it is a hub of various AI models, datasets, and live AI demos (referred to as spaces in Hugging Face jargon). Hugging Face is quickly evolving into an AI community where developers are sharing cutting-edge AI work and having discussions that push the frontier of AI.
Exploring Hugging Face components relevant to PyTorch
We can view Hugging Face as a platform that contains various components, as shown in Figure 19.1. You can access the page shown in the figure on the Hugging Face website [2]. The first thing to note in this figure is the mention of PyTorch in various Hugging Face components. Hugging Face’s libraries, models, and datasets are fully compatible with PyTorch, so it is wise to discuss Hugging Face in detail...