Could anyone clarify whether we can import private utility scripts into our submission notebooks during the evaluation stage? I can't see anything specifically precluding their usage, but would be very grateful if someone can confirm that they are allowed!
", "messages": [ { "id": 1668745, "postDate": "2022-01-30T01:37:14.910Z", "content": "Yes I would say for sure, as long as your are not breaking license agreements or something like that.
", "rawMarkdown": "Yes I would say for sure, as long as your are not breaking license agreements or something like that.", "votes": 1, "replies": [ { "id": 1669064, "postDate": "2022-01-30T10:10:40.410Z", "content": "Thanks for replying Chris. Had a scan around and couldn't see any mention in the rules - seemed like Kaggle is treating them a bit like a dataset so was wondering whether they needed to be made public for that reason. Have you used them successfully in competitions (e.g. Optiver) before?
", "rawMarkdown": "Thanks for replying Chris. Had a scan around and couldn't see any mention in the rules - seemed like Kaggle is treating them a bit like a dataset so was wondering whether they needed to be made public for that reason. Have you used them successfully in competitions (e.g. Optiver) before?", "votes": 1 }, { "id": 1669491, "postDate": "2022-01-30T18:19:10.810Z", "content": "Yes I have. You can also save your models to a private dataset and use that. If they every want you to run all of your code from scratch in one notebook they will make it clear. Otherwise, you can always use utility scripts to organize your code.
", "rawMarkdown": "Yes I have. You can also save your models to a private dataset and use that. If they every want you to run all of your code from scratch in one notebook they will make it clear. Otherwise, you can always use utility scripts to organize your code. ", "votes": 1 }, { "id": 1669508, "postDate": "2022-01-30T18:39:06.703Z", "content": "Hmm, ok - I thought that any datasets had to be made public though? My submissions in this competition have failed when they have depended on my private datasets, reporting the error Notebook Exceeded Allowed Compute. This isn't a RAM issue or the like. Even having the dataset associated with my notebook caused the submission to fail even when submitting rapid-fire zero predictions. The rules seemed to confirm this:
\n\n\n[7] C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants.
\n
For private utility scripts the submissions are passing however - hence my uncertainty.
", "rawMarkdown": "Hmm, ok - I thought that any datasets had to be made public though? My submissions in this competition have failed when they have depended on my private datasets, reporting the error Notebook Exceeded Allowed Compute. This isn't a RAM issue or the like. Even having the dataset associated with my notebook caused the submission to fail even when submitting rapid-fire zero predictions. The rules seemed to confirm this:\n\n> [7] C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants. \n\nFor private utility scripts the submissions are passing however - hence my uncertainty." }, { "id": 1669509, "postDate": "2022-01-30T18:39:22.857Z", "content": "Any chance you could clarify please, @sohier ?
", "rawMarkdown": "Any chance you could clarify please, @sohier ?" }, { "id": 1670748, "postDate": "2022-01-31T20:39:04.963Z", "content": "The same rules apply to private utility scripts as private datasets. If the contents are your own work you can leave them private, etc.
", "rawMarkdown": "The same rules apply to private utility scripts as private datasets. If the contents are your own work you can leave them private, etc.", "votes": 2 }, { "id": 1670777, "postDate": "2022-01-31T21:14:09.060Z", "content": "Great, that makes things easier. I didn't realise that the provenance of the contents mattered - thanks for clarifying. Thanks also @chrisrichardmiles for your input.
", "rawMarkdown": "Great, that makes things easier. I didn't realise that the provenance of the contents mattered - thanks for clarifying. Thanks also @chrisrichardmiles for your input." } ] }, { "id": 1670773, "postDate": "2022-01-31T21:04:40.777Z", "content": "Nvm, it's working now.
", "rawMarkdown": "~~I have built a python package and generated a model for this competition and planned to use a private dataset for submission but the kernel does not allow me to use private or public data sets when internet is turned off. Not sure what I'm doing wrong?~~ Nvm, it's working now. " }, { "id": 1668669, "postDate": "2022-01-29T21:44:12.463Z", "content": "Could anyone clarify whether we can import private utility scripts into our submission notebooks during the evaluation stage? I can't see anything specifically precluding their usage, but would be very grateful if someone can confirm that they are allowed!
", "rawMarkdown": "Could anyone clarify whether we can import private utility scripts into our submission notebooks during the evaluation stage? I can't see anything specifically precluding their usage, but would be very grateful if someone can confirm that they are allowed!" }, { "id": 1670768, "postDate": "2022-01-31T20:58:36.490Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1668745, "author_name": "Chris Miles", "author_url": "", "post_date": "2022-01-30T01:37:14.910000", "content": "Yes I would say for sure, as long as your are not breaking license agreements or something like that.
", "votes": 1, "replies": [ { "id": 1669064, "author_name": "jagofc", "author_url": "", "post_date": "2022-01-30T10:10:40.410000", "content": "Thanks for replying Chris. Had a scan around and couldn't see any mention in the rules - seemed like Kaggle is treating them a bit like a dataset so was wondering whether they needed to be made public for that reason. Have you used them successfully in competitions (e.g. Optiver) before?
", "votes": 1, "replies": [] }, { "id": 1669491, "author_name": "Chris Miles", "author_url": "", "post_date": "2022-01-30T18:19:10.810000", "content": "Yes I have. You can also save your models to a private dataset and use that. If they every want you to run all of your code from scratch in one notebook they will make it clear. Otherwise, you can always use utility scripts to organize your code.
", "votes": 1, "replies": [] }, { "id": 1669508, "author_name": "jagofc", "author_url": "", "post_date": "2022-01-30T18:39:06.703000", "content": "Hmm, ok - I thought that any datasets had to be made public though? My submissions in this competition have failed when they have depended on my private datasets, reporting the error Notebook Exceeded Allowed Compute. This isn't a RAM issue or the like. Even having the dataset associated with my notebook caused the submission to fail even when submitting rapid-fire zero predictions. The rules seemed to confirm this:
\n\n\n[7] C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants.
\n
For private utility scripts the submissions are passing however - hence my uncertainty.
", "votes": 0, "replies": [] }, { "id": 1669509, "author_name": "jagofc", "author_url": "", "post_date": "2022-01-30T18:39:22.857000", "content": "Any chance you could clarify please, @sohier ?
", "votes": 0, "replies": [] }, { "id": 1670748, "author_name": "Sohier Dane", "author_url": "", "post_date": "2022-01-31T20:39:04.963000", "content": "The same rules apply to private utility scripts as private datasets. If the contents are your own work you can leave them private, etc.
", "votes": 2, "replies": [] }, { "id": 1670777, "author_name": "jagofc", "author_url": "", "post_date": "2022-01-31T21:14:09.060000", "content": "Great, that makes things easier. I didn't realise that the provenance of the contents mattered - thanks for clarifying. Thanks also @chrisrichardmiles for your input.
", "votes": 0, "replies": [] } ] }, { "id": 1670773, "author_name": "Brayan Jaramillo", "author_url": "", "post_date": "2022-01-31T21:04:40.777000", "content": "Nvm, it's working now.
", "votes": 0, "replies": [] }, { "id": 1670768, "author_name": "", "author_url": "", "post_date": "2022-01-31T20:58:36.490000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1668745": "Yes I would say for sure, as long as your are not breaking license agreements or something like that.", "1670773": "~~I have built a python package and generated a model for this competition and planned to use a private dataset for submission but the kernel does not allow me to use private or public data sets when internet is turned off. Not sure what I'm doing wrong?~~ Nvm, it's working now. ", "1668669": "Could anyone clarify whether we can import private utility scripts into our submission notebooks during the evaluation stage? I can't see anything specifically precluding their usage, but would be very grateful if someone can confirm that they are allowed!", "1670768": "" } }