[codesushi42]

VSCode Intellisense also uses a deep learning approach for autocompletion.

How does this compare?

[jacob-jackson]

As far as I can tell, it suggests one token at a time and uses its model to help rank these tokens. This is useful, but there is a lot to be gained by suggesting multiple tokens at once.

TabNine has always included a logistic regression model to help rank completions. It uses features such as the occurrence frequency of the token and the number of similar contexts in which it occurs.

[codesushi42]

Intellisense also outputs multiple suggestions.

Also, TabNine mentions they are using transformers, which is not logistic regression. The context will be inferred using attention.

[jacob-jackson]

Sorry, I worded that poorly. What I mean is that each individual suggestion consists of a single token -- at least, that is what I see in https://visualstudio.microsoft.com/services/intellicode/. Compare the videos in https://tabnine.com/blog/deep, where most suggestions consist of multiple tokens.

Deep TabNine (announced today) uses transformers. TabNine (released last year) uses logistic regression.

[codesushi42]

Interesting, thanks.