Artificial Intelligence: What Are Large Models?

by

Welcome to the special winter vacation column “High-Tech Lessons for Kids” launched by Science Popularization China!

As one of the cutting-edge technologies today, artificial intelligence is changing our lives at an astonishing pace. From smart voice assistants to self-driving cars, from AI painting to machine learning, it opens up a future full of infinite possibilities. This column will explain the principles, applications, and profound impacts of artificial intelligence on society in a simple and understandable way, using videos and text for children.

Let’s embark on this AI journey together!

First, let’s watch the video:

Below is the text version:

(Reading takes about 1 minute)

Recurrent Neural Networks

In everyday language, large models generally refer to large language models. The meaning of large language models is easy to understand; they are models trained on a vast amount of language text data to understand and generate human language.

The amount of data used to train large language models and the number of parameters in these models are both very large.

For example, in 2018, the dataset for training GPT-1 contained approximately close to 1 billion words. At that time, the BERT model was trained on 3.3 billion words. In 2022, the dataset used to train GPT-3.5 exceeded 45TB, and the GPT model contains over 100 billion parameters.

With such a large number of samples and parameters, large models demonstrate better text understanding and reasoning abilities than ordinary models, allowing them to better understand and answer the questions we pose.

However, due to the need for a large amount of data and extensive computations, the training cost of large models is extremely high. The training cost for a year can reach millions of RMB. Therefore, there are relatively few companies with sufficient economic strength to create large models.

Currently, many companies claim to be developing their own large models, but in reality, they may not qualify as true large models.

Planning and Production

This article is a work of the Science Popularization China – Creative Cultivation Program.

Produced by | Science Popularization Department of China Association for Science and Technology

Supervised by | China Science and Technology Publishing House Co., Ltd., Beijing Zhongke Xinghe Cultural Media Co., Ltd.

Author | Beijing Yunyuji Cultural Communication Co., Ltd.

Reviewed by | Qin Zengchang, Associate Professor, School of Automation Science and Electrical Engineering, Beihang University

(Science Popularization China)

Artificial Intelligence: What Are Large Models?

Leave a Comment