Natural Language Processing

• Transformer Models are programmable. You can provide them with enough context (i.e., Memory), possibly from a data store (i.e., Virtual Memory). They then perform some processing and give you a response. For more info on prompt engineering see this

• Transformers are so effective because learning from human language is a cheat. Human language encodes human thinking, and transformers learn to exploit this.

Topics §

• Sequence Based Algorithms
• Transformer Model
• Language Model
• Large Language Model

Links §

• Speech and Language Processing - An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition with Language Models by Jurafsky and Martin