| Developers: | Yandex |
| Date of the premiere of the system: | June 2022 |
| Branches: | Information Technology |
2022: Placing the neural network in the public domain
In June 2022, Yandex posted a neural network for generating texts in Russian and English. The company calls its development (it was called YaLM 100B) the largest GPT-like model, which is available to everyone.
Language models from the YaLM family determine the principle of its construction from the finished text and generate a similar one. For example, they can come up with ideas for advertising campaigns, create descriptions of goods and videos, generate poems, answers, congratulations and classify them by speech style and other parameters.
The model was trained on Yandex supercomputers. YaLM 100B processed about 2 TB of texts in English and Russian from open datacets and the Internet.
It is noted that the neural network contains 100 billion parameters and is the largest of the existing models for the Russian language. This allows you to use it to solve a large range of problems related to natural language processing. Language models from the YaLM family determine the principle of text construction and generate new ones based on the laws of linguistics and their knowledge of the world, Yandex said in a statement.
For example, language models are able to come up with, in particular, ideas for advertising campaigns, create descriptions of goods and videos. With their help, you can generate any texts, as well as classify them, for example, by speech style.
Pyotr Popov, CEO of Yandex Technologies, said that by putting YaLM 100B in the public domain, the company expects that this will give an impetus to the development of generative neural networks. The model is provided under an open license Apache 2.0 and is available on GitHub.[1]
By June 2022, Yandex already uses YaLM neural networks in more than 20 projects, including Yandex. Search, "voice assistant" Alice, "they also help support staff respond to calls, generate advertisements and site descriptions.[2]
Notes
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |