Now programming can be made easier, thanks to a new tool. It's called Deep TabNine, a new auto-completion tool based on deep learning. It suggests complete lines of code based on what you've written so far and works with multiple programming languages.
The tool was created by a student at the University of Waterloo in Canada, Jacob Jackson, and is based on the GPT-2 language model from Open AI. Deep TabNine aims to help make programming a faster process.
Automatic code completion via deep learning
Deep TabNine is a deep learning-based tool, which was trained on two million files from GitHub.
Just as GPT-2 was trained in the same way, to predict sentences, Deep TabNine predicts the building blocks of code.
Deep TabNine supports many languages such as Java, Python, JavaScript, C, C++, PHP, TypeScript, Kotlin, Objective-C, HTML, CSS, Go, C#, Ruby, Rust, Swift, Haskell, OCaml, Scala, Perl, SQL, and Bash.
There are of course other tools like Deep TabNine already available on the market, including Microsoft for Visual Studio. However, Deep TabNine's ability to suggest multiple completion lines instead of just one is what makes this autocomplete tool so special.
A standard version of TabNine is also available, but it does not rely on deep learning. It uses machine learning to provide reliable and relevant suggestions.
However, it should be noted that the new auto‑completion tool comes with a drawback, as it is very “heavy” to run on a laptop and may not provide suggestions as quickly as the standard version of TabNine.
Jackson hopes to build a better model that can easily run on laptops. For now, he's offering a beta service called TabNine Cloud that uses GPUs to speed up autocomplete suggestions.
In the meantime, he/she is working on a model that will run on a laptop with “logical delay”. For corporate clients, the tool offers a license to use the model for the company's hardware.
