I am currently working on my MSc thesis, where I try to automatically convert description of an algorithm in natural language to source code in Python (the algorithms are quite simple, like revert an array of 100 elements) using deep learning. The main problem is that I need a lot of data to do that. Does anyone know any available datasets consisting of pairs {short natural language description, source code}?

I know the Heartstone cards dataset (really useful and close to my needs, but still not enough), the Django dataset (Django code commented line by line - it doesn't really contain the description of whole algorithm, it rather translates english to Django code line by line). I tried to contact with few sites like for eg. Sphere Online Judge but to no avail.

Every help would be appreciated.


1 Answer 1


StackOverflow answers with code snippets. This data needs some processing, because the description can be in the question (along with other notes) and along with the answer. But this dataset is very big.

Also take a look at CodeReview questions.


Not the answer you're looking for? Browse other questions tagged .