Sergey Vilov

AI

Constrained Sentence Generation Using Gibbs Sampling and BERT

A fast and effective approach to generating fluent sentences from given keywords using public pre-trained models Sergey Vilov · Follow Published in Towards Data Science · 10 min read · 9 hours ago — Photo by Brett Jordan on Unsplash Large language models, like GPT, have achieved unprecedented results in free-form text generation. They’re widely used for writing e-mails, copyrighting, or storytelling. However, their success in constrained text generation remains limited [1]. Constrained text generation

Read More »