Deep Learning#

Most modern NLP are language models. While a language model can be anything that predicts new text, in practice, these models are based on deep neural networks. In these notebooks, we will start from the very basics of training a language model from scratch and build up to training much larger and more advanced architectures. Importantly, this material is very hard and not intended, as of now, to be comprehesive. Instead, I recommend you work through these notebooks alongside other deep learning resources. For example, Jeremy Howard’s Practical Deep Learning for Coders is incredibly helpful. The bottom line is that no one was born knowing deep learning and that it can help to see this material presented in several different ways.