The AI model uses Lu Xun's style to write the composition of the college entrance examination. How many points can everyone come to Pinpin?

Yun Zhong sent from Aofei Temple
Qubit | Public Account QbitAI

As soon as the composition topic of the college entrance examination came out, it once again became the focus of people's attention.

And this time there were some AI large-scale model candidates, and their performance also attracted crowds of onlookers.

For example, Xiao Yuan, who has read 3 million books, has read extensively, and is full of knowledge, came to challenge the composition of the college entrance examination!

583f9e5ba4718aea9a4ebef8360013af.gif

Not long ago, Xiao Yuan incarnated as Lu Xun to have a dialogue with young people. Today, Xiao Yuan incarnates as Lu Xun again to challenge the National College Entrance Examination Paper A:

29419df4301ea4d6dce0bec95b78aa0c.png

Let's take a look at how many points Xiao Yuan can get for this composition? (full score 60)

How do we see technological innovation now

268bbc66a943f2b0392f1f45fc22027c.png

ca125d3198a42bc7641184b83e4abbc3.png73e6d56a9aef6b4a6fc49e2a84a4b5e0.png

Xiaoyuan who has read 3 million books

The source 1.0 massive Chinese AI model  (hereinafter referred to as "source 1.0") is a single massive natural language processing model released by Inspur Information in September 2021. When it was released, it became the largest Chinese AI model in the world. Its parameter scale Up to 245.7 billion , the high-quality Chinese data set used for training reaches 5000GB .

"Yuan 1.0" has excellent performance in language intelligence. When it was released, it won the championship of zero-sample learning and small-sample learning in the Chinese language comprehension evaluation benchmark CLUE list; at the same time, it reached the top of the Chinese language comprehension and generation evaluation benchmark CUGE (Intellectual Source Index) ranked first in the simplified list, and won the best results in five evaluations including language comprehension (text level), language generation, dialogue interaction, multilingualism, and mathematical reasoning.

With its excellent small sample learning ability, Yuan 1.0 can adapt to the needs of various AI tasks. Based on the Yuan 1.0 model, a variety of different AI applications such as intelligent customer service robots, copywriting, intelligent content processing, and virtual digital humans have been launched.

Reprinted from Jack-cui, the UP master of station B

Alright, let's evaluate objectively. How can Xiaoyuan score this performance?

Guess you like

Origin blog.csdn.net/QbitAI/article/details/131118330