Further release and details
Great work for your guys! I was wondering when the rest of Refinecode would be released.
Besides, which part of the Stack V2 do you use to cook Refinecode?
To use this metainfo, I guess, I should first download the stack V2 datasets and use its blob-ids to get the file contents from aws S3. However, the full version is too large (> 1TB) to download and use.
So I want to make sure which version your "Stack V2" refers to: bigcode/the-stack-v2-dedup or bigcode/the-stack-v2 or bigcode/the-stack-v2-train-full-ids?
Or would you kindly give me some suggestions on how to use this dataset more efficiently?
Hi, we are still working hard to release the rest part. It won't be a long time.
We use bigcode/the-stack-v2-dedup as a part of original code source. However, due to our data processing pipeline, it is hard to provide the corresponding blob-ids. We suggest to obtain those files from Github by using the given repo_name and file_path, as it is easy to crawl raw code content from Github directly.
It is hard to crawl raw code in this way, hh
Hi, we are still working hard to release the rest part. It won't be a long time.
We use bigcode/the-stack-v2-dedup as a part of original code source. However, due to our data processing pipeline, it is hard to provide the corresponding blob-ids. We suggest to obtain those files from Github by using the given repo_name and file_path, as it is easy to crawl raw code content from Github directly.
Thanks for this suggestion!
I tried to download the files from GitHub, but unfortunately, the rate limit makes this impossible: https://docs.github.com/en/rest/using-the-rest-api/rate-limits-for-the-rest-api