Learn the principles behind big language models: The breakthrough reason behind ChatGPT, how 25,000 computers were trained

Please add a picture description

What word comes after Good?

You might think "good morning" or "goodbye". But you definitely don't say "Good Loud". It doesn't make sense at all. Computer scientists have been training artificial intelligence to solve this problem for decades.

Please add a picture description
Given a word, our AI predicts the next word. Do this a few times and you've generated a sentence.

This is how ChatGPT works.

ChatGPT has been trained across the internet and learned how to chat like a human. However, this monumental feat was only possible with a breakthrough in the late 2010s. Underpinning ChatGPT's breakthroughs and forever shaping the world we live in.

This is the story of an artificial intelligence that reads and learns from every book, tweet and website on the entire internet. and how it is possible.

Sentences are long.

When we go beyond single words, predicting the next word becomes more difficult. Take this example.

Please add a picture description
Saying "I had a good morning" doesn't make sense in this situation. But our AI only sees the good and spits out morning. In most cases, even humans need many words to predict the next word. Therefore, artificial intelligence also needs this additional information.

Our AI needs to read many words before it can predict the next word. ChatGPT can read more than 8,000 previous words at a time. The natural way to do this is to assign each

Guess you like

Origin blog.csdn.net/iCloudEnd/article/details/132624187