Методи синтезу надшвидкодiючих структур мовних систем штучного iнтелекту

Abstract

This work has goal to address the current approaches of implementing superfast language syntax parsers. It contains general overview of modern solutions based on neural networks and a more in depth description was provided for a particular model that is implemented using transformer networks – BERT. This includes providing benchmarks to understand its current capabilities and field where it can be possibly used