英伟达NVIDIA：2024年大语言模型（LMM）新手入门指南第一部分（英文版）

The switch from one methodology to another was largely driven by relevant technological and methodological advancements, such as the advent of neural networks, attention mechanisms, and transformers and developments in the field of unsupervised and self-supervised learning. The following sections will briefly explain these concepts, as understanding them is crucial for truly understanding how LLMs work and how to build new LLMs from scratch.