A large model refers to a deep learning model with a very large number of parameters, typically used to process large-scale datasets and capable of learning complex features of the data, thus having significant advantages in handling complex tasks. Large scale models have gone through three stages of development: pre trained models, large-scale pre trained models, and ultra large scale pre trained models, with parameter scales ranging from billions to trillions. The Transformer architecture has become the mainstream algorithm foundation in the field of large models, giving rise to two main technical routes: GPT and BERT. The modes supported by the large model are more diversified, from a single task under a single mode to supporting multiple tasks under multiple modes. The application areas of big models include general big models and industry big models. According to input types, there are three main categories: atmospheric language models (NLP), visual models (CV), and multimodal models. The core principle of the big model is based on deep learning technology, which involves processing massive amounts of data.
