FROM OUR BLOG

The transformer-based language models have been showing promising progress on a number of different natural language processing (NLP) benchmarks. The combination of transfer learning methods with large-scale transformer language models is becoming a standard in modern NLP. In this article, we will make the necessary theoretical introduction to transformer architecture and text classification problem. Then we will demonstrate the fine-tuning process of the pre-trained BERT model for text classification in TensorFlow 2 and Pytorch.

AI & machine learning is becoming part of today's products more than ever before. It is now even more important for each business to have a clear vision of how artificial intelligence intersects with their mission. This is the list what we recommend to our partners to get insights on current ecosystem.

GraphQL is a query language for building API and a server-side runtime, which enables you to describe your data. GraphQL isn't tied to any specific database or programming language. You can build your GraphQL server in different languages like Node.js, C#, Scala or Python. The ecosystem in Node.js and JavaScript in general is the most evolved GraphQL wise and it is usually better to build API for client facing application in Node.js. You can take a look at our blog to learn more. However, Node.js is not the best fit for data science, so we will take a look at how to build microservices that will help us serve our machine learning models, high-performance computations and text analytics. In this article, we will be solely focused on building an API that enables us to expose text analytics capabilities. Our GraphQL server can be also used to serve the output from machine learning models built by frameworks like TensorFlow, FastText or PyTorch, but we will take a look at this in depth in our future articles.

The React Spring library has a modern way of approaching animations by using a physics-based model. You can configure animations using real-life values like mass, friction, and tension, which allow you to create more believable transitions. There is also an option to set the duration in milliseconds. The library has great performance and an easy-to-use API. React Spring is cross-platform and has quite a optimal bundle size.

Figma, the platform that forced even the most conservative designers to switch from Sketch, Invision or AdobeXD, is here with new improvements. Until recently, plugins were something that we really missed in Figma in comparison with other platforms. In the last few months, Figma has launched many updates such as GIF import or Smart Selection, so we expected that plugins in Figma would finally be introduced in the near future. Now it has almost been three month since the release and there are over 150 available plugins, so I have collected the 7 most useful plugins that will enable you to make your designs much better and faster than ever before.

The hype following the last React conference has diminished. The proposal for React Hooks was introduced as part of the React alpha release. Since React v16.8, React Hooks have been in all official releases as well. How does this improvement affect how we build our apps with GraphQL?