Sber has released its own version of the multimodal neural network
[ad_1]
Sber is launching its own GigaChat multimodal neural network in test mode, at the first stage access to it will be provided by invitation. This was reported by the press service of Sberbank.
The architecture of the service, as noted, is based on the neural network ensemble of the NeONKA (NEural Omnimodal Network with Knowledge-Awareness) model, which includes various neural network models. Due to this, the neural network can support a conversation, generate texts and images, and write program code.
The model is already being implemented in many Sber services, the first of which, in which all its capabilities will be implemented in full, will be the Salyut virtual assistant.
Against the backdrop of the popularity of ChatGPT and the neural network for generating Midjourney images from August 2022 to February 2023, the number of users of neural networks in Russia has grown 5 times, and the duration of interaction with AI – 3 times, previously calculated by Tele2 analysts.
The rise in popularity of GPT chats has sparked a debate on several fronts in the media and the internet. We are talking about both the change in the labor market due to digitalization, and the vulnerability of certain categories – for example, specialists associated with the accumulation and systematization of knowledge (scientists, lawyers), or representatives of creative professions.
[ad_2]
Source link
