Yandex published to open access to the new YandexGPT 5 Lite Predrain language model, said the company’s press service to RB.RU. This is a neuronal network with 8 billion parameters capable of processing the context of up to 32 thousand tokens (approximately 24 thousand words). It is available for free in Hugingface.
Author:
https://rb.ru/author/yuliya-pazhitnyh/
Subscribe to RB.ru on Telegram
“[Нейросеть] It will be useful for developers and commercial representatives who can receive it under their tasks and not waste time creating their own basic model, ”said the company. They added that the last time they placed the language model in open access in 2022.
The language model was taught in two stages. First, it was trained in texts, code, calculations, calculations and other Russian and English -speaking data with a volume of 15 billion tokens. In the second stage, “high quality data” were used, including texts from an educational topic with a volume of 320 billion tokens, the company said.
- YandexGPT 5 Lite Pretrain is a representative of the new generation of Yandex neuronal networks along with a more advanced Yandexgpt 5 pro. The company previously told RB.RU that they integrated Yandexgpt 5 Pro in Alice, can access the model in the Yandex, Mobile and Desktop versions of the Yandex browser, as well as on intelligent devices.
- According to the company, the new model is comparable in the quality of the responses with the GPT-4O of OpenAI and in 64% of the cases the Chinese model QWEN2.5 exceeds to solve standard problems.
Author:
Julia Pastor
Source: RB
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
