Sber has introduced the multimodal neural network GigaChat, which looks like a real Russian competitor to the American ChatGPT. The artificial intelligence, built on the neural network model NeONKA (NEural Omnimodal Network with Knowledge-Awareness), answers user questions, writes program code, and even draws pictures based on text descriptions. Sber emphasizes that GigaChat communicates more competently in Russian than ChatGPT. Which, in general, is not surprising.
The NeONKA model is already being implemented in many Sber products and services: the virtual assistant Salut will be one of the first products with all the capabilities of the new model. The company also announced that ru-GPT 3.5, which is part of NeONKA with 13 billion parameters, will be made publicly available "in order to accelerate the development of the open-source community and technologies in the field of artificial intelligence." The model is currently running in a closed beta version.
Now on home
Vladislav Antipov: Polymer materials for the wide-chord blade will reduce the weight of the unit
Geomagnetic disturbances of level G1-G2 are a thing of the past, background indicators have returned to the green zone
"Progress MS-33" Flies to Replace it with Cargo for the Station
Solikamsk Magnesium Plant uses materials with neodymium, cerium, and lanthanum
Andrey Dutov: the founder of a startup must have an understanding of who he will sell the idea to
Power units are exposed to crystalline ice
The device protects the primary circuit from pressure surges during normal operation and emergency situations
Vietbao: the Checkmate project will ensure the economic viability of fighters
The product combines high autonomy and accuracy
The Ministry of Internal Affairs will have the right to block operators for fake diagnostic cards without the participation of insurers from September