Text Classification datasets and models for Uktainian

We release datasets and models for Ukrainian covering several classification domains: toxicity, NLI, and formality.