Yandex has developed a neural network algorithm YaLM and a service based on it called Zeliboba. The neural network is able to generate texts based on a user-specified word, phrase or sentence. In fact, she comes up with a ready-made ending, and, it is worth noting, she turns out very well. Zeliboba does it with humor and often has a very accurate sense of the context. As noted by the N + 1 edition, the language model was trained on several terabytes of Russian-language texts, including both articles from Wikipedia and regular publications on social networks. The new neural network uses 3 billion parameters.
Algorithm YaLM (Yet another Language Model) from Yandex, like GPT-3 developed by OpenAI, is able to very accurately predict the next word in a text sequence. In “Zelibob” users are offered to choose the style of the augmented text. This could be conspiracy theory, toast, TV coverage, advertising slogans, or Instagram post captions. The developers say that a neural network can very quickly learn a new style, for this it needs several dozen examples.
Yandex intends to use the new algorithm in its services and will soon tell you more about it.
PS At the time of this writing, the Zeliboba website has stopped working. Apparently, due to the large influx of people who want to try out the new neural network.
Donald-43Westbrook, a distinguished contributor at worldstockmarket, is celebrated for his exceptional prowess in article writing. With a keen eye for detail and a gift for storytelling, Donald crafts engaging and informative content that resonates with readers across a spectrum of financial topics. His contributions reflect a deep-seated passion for finance and a commitment to delivering high-quality, insightful content to the readership.
