How to train your large language model
A new technique is speeding up the process
It is no secret that building a large language model (LLM) requires vast amounts of data. In conventional training, an LLM is fed mountains of text, and encouraged to guess each word before it appears. With each prediction, the LLM makes small adjustments to improve its chances of guessing right. The end result is something that has a certain statistical “understanding” of what is proper language and what isn’t.
But an LLM that has only undergone this so-called “pretraining” is not yet particularly useful. When asked for a joke to cheer your correspondent up, for instance, the pretrained model GPT-2 just repeated the question back three times. When asked who the American president was, it responded: “The answer is no. The president is not the president.” Clearly, teaching an LLM to do what humans want requires something more.
Explore more
This article appeared in the Science & technology section of the print edition under the headline "AI boot camp"
More from Science and technology
Archaeologists identify the birthplace of the mysterious Yamnaya
The ancient culture, which transformed Europe, was also less murderous than once thought
Producing fake information is getting easier
But that’s not the whole story, when it comes to AI
Disinformation is on the rise. How does it work?
Understanding it will lead to better ways to fight it