OpenAI: GPT-3 language model generates 4.5 billion words a day

The GPT-3 language model, developed by the nonprofit OpenAI, generates about 4.5 billion words a day, according to its developers.

\n\n\n\n

OpenAI also noted that the model achieved these results thanks to more than 300 applications and tens of thousands of developers.

\n\n\n\n

For example, the startup Viable uses GPT-3 to analyse customer feedback, identifying emotions and sentiments from surveys, support tickets, or reviews.

\n\n\n\n

Fable Studio uses OpenAI’s language model to create dialogues in virtual reality, while Algolia uses it to improve search products.

\n\n\n\n

According to some analysts, basing a startup on someone else’s technology is a bad idea: the ease of using GPT-3 to create a product will increase the number of competitors. Yet no company will benefit from this technology as much as OpenAI itself, skeptics say.

\n\n\n\n

Experts also see benefits for Microsoft, which in September 2020 получила эксклюзивное право на использование языковой модели на облачной платформе Azure.

\n\n\n\n

As reported in January 2021, OpenAI показала a GPT-3-based algorithm called DALL-E, which, from a textual description, generates realistic images of non-existent things.

\n\n\n\n

Earlier Google представила свою языковую модель с 1 трлн параметров, что в шесть раз превысило показатель GPT-3.

\n\n\n\n

Subscribe to ForkLog news on VK!

\n