Kozhevnikov Artyom@Osborn

Protecting users from hacker attacks. Most hacker attacks on companies happen because of the human factor. Our solution protects all employees from fraudulent attacks. It can be applied to an ordinary person making purchases or sales as well as to a large company. We use to replace phones, e-mails and other confidential information with special tokens to protect companies from data leaks. And that would eliminate model bias to certain data. To make the model work, you only need to generate one token, which allows you to use the model with minimal effort. And a system to get a percentage of confidence based on logprobs lets you know how much of the prediction is correct.

This event we will focus on GPT-3 functions such as Files, Answers & Fine-tuning