![\"settings](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F473234%2F4fa20f9fbba0b237d4d047554ce7b28c%2Fnotebook_settings.png?generation=1595019964465999&alt=media\")
Hi.\nI'm confused about custom packages.
\n\nCode Requirements says \"No custom packages enabled in kernels\". I think this means that we participants are not allowed to use custom packages in submission notebooks.
\n\nI use some python packages not included in Kaggle Notebooks for training but don't use them in submission notebook. (If some parts are necessary, I code them.)
\n\nSome public notebooks, however, succeeded in submission though those used codes or packages which are uploaded as Kaggle Dataset.
\n\nMy question is simple, what \"custom packages\" means ?
\n\nThanks.
", "messages": [ { "id": "920421", "postDate": "07/08/2020 15:33:38", "content": "Hi.\nI'm confused about custom packages.
\n\nCode Requirements says \"No custom packages enabled in kernels\". I think this means that we participants are not allowed to use custom packages in submission notebooks.
\n\nI use some python packages not included in Kaggle Notebooks for training but don't use them in submission notebook. (If some parts are necessary, I code them.)
\n\nSome public notebooks, however, succeeded in submission though those used codes or packages which are uploaded as Kaggle Dataset.
\n\nMy question is simple, what \"custom packages\" means ?
\n\nThanks.
", "rawMarkdown": "Hi.\nI'm confused about _custom packages_.\n\n[Code Requirements](https://www.kaggle.com/c/birdsong-recognition/overview/code-requirements) says _\"No custom packages enabled in kernels\"_. I think this means that we participants are **not allowed** to use custom packages in **_submission notebooks_**. \n\nI use some python packages not included in Kaggle Notebooks for **_training_** but don't use them in **_submission notebook_**. (If some parts are necessary, I code them.)\n\nSome public notebooks, however, **succeeded in submission** though those used codes or packages which are uploaded as Kaggle Dataset.\n\nMy question is simple, what _\"custom packages\"_ means ?\n\nThanks.", "votes": null }, { "id": "921143", "postDate": "07/09/2020 05:18:22", "content": "I believe that since internet needs to be disabled in the submission notebook, it is not possible to install new packages. However I also believe that as kaggle staff sees common use of some package, they add it to the new versions of the docker
", "rawMarkdown": "I believe that since internet needs to be disabled in the submission notebook, it is not possible to install new packages. However I also believe that as kaggle staff sees common use of some package, they add it to the new versions of the docker", "votes": null }, { "id": "921204", "postDate": "07/09/2020 06:02:21", "content": "I think they meant some package which you made up locally and is not available to use publically.I am not sure though.
", "rawMarkdown": "I think they meant some package which you made up locally and is not available to use publically.I am not sure though.", "votes": null }, { "id": "921208", "postDate": "07/09/2020 06:04:09", "content": "It is possible to install packages without internet.
", "rawMarkdown": "It is possible to install packages without internet.", "votes": null }, { "id": "921743", "postDate": "07/09/2020 14:10:50", "content": "Here is a similar question that was asked in the Toxic tweet competition: discussion
", "rawMarkdown": "Here is a similar question that was asked in the Toxic tweet competition: [discussion](https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification/discussion/88021)", "votes": null }, { "id": "921951", "postDate": "07/09/2020 17:18:26", "content": "Thank you for information !!
\n\nI did not participate in the competition and missed the topic.
", "rawMarkdown": "Thank you for information !!\n\nI did not participate in the competition and missed the topic.", "votes": null }, { "id": "921978", "postDate": "07/09/2020 17:40:12", "content": "Thank you for reply. \nSo you mean that we can use package (or codes) such as open GitHub Repository which anyone can use ?
", "rawMarkdown": "Thank you for reply. \nSo you mean that we can use package (or codes) such as open GitHub Repository which anyone can use ?", "votes": null }, { "id": "930659", "postDate": "07/15/2020 16:17:55", "content": "This is a great question. Saying \"no custom packages enabled\" is meaningless in python, especially when \"custom datasets\" are enabled. A package is just a bunch of code files so what if I upload my own python files ? Are they custom packages ? Are they allowed ? I'm also confused !
", "rawMarkdown": "This is a great question. Saying \"no custom packages enabled\" is meaningless in python, especially when \"custom datasets\" are enabled. A package is just a bunch of code files so what if I upload my own python files ? Are they custom packages ? Are they allowed ? I'm also confused !", "votes": null }, { "id": "930891", "postDate": "07/15/2020 19:38:06", "content": "Yes, I have the same question, too.\nIn the previous code competition, I uploaded my python code as a dataset and used it because the rule \"no custom packages enabled\" didn't exist.
", "rawMarkdown": "Yes, I have the same question, too.\nIn the previous code competition, I uploaded my python code as a dataset and used it because the rule \"no custom packages enabled\" didn't exist.", "votes": null }, { "id": "931148", "postDate": "07/16/2020 02:45:37", "content": "yes,I think so.
", "rawMarkdown": "yes,I think so.", "votes": null }, { "id": "933497", "postDate": "07/17/2020 18:19:43", "content": "@addisonhoward can you please comment on this topic!
", "rawMarkdown": "addisonhoward can you please comment on this topic!", "votes": null }, { "id": "933546", "postDate": "07/17/2020 19:07:35", "content": "No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed. If you try to submit an output from a kernel with this enabled, you will be blocked on submission.
\n\nUsing code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.
", "rawMarkdown": "No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed. If you try to submit an output from a kernel with this enabled, you will be blocked on submission.\n\nUsing code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.", "votes": null }, { "id": "933663", "postDate": "07/17/2020 21:52:38", "content": "@addisonhoward \nThank you for reply.\n
\n> No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed.
I think one source of confusion is that we cannot find “Packages” in kernel's \"Settings\". (Or you mean another \"Settings\"? )
\n\n
> Using code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.
\n\nI understand we can use public codes, pretrained model, external data by attaching them as dataset if and only if they are declared.
\nOn the other hand, in order to use our own (not public) code such as a custom model definition, we must code it on notebook or using utility script, right ?
\nCorrect me if I misunderstand. Thanks.
My apologies.
\n\nThis was referencing a function that was previously available in Kaggle Notebooks/Kernels but we have since deprecated. We have updated the code requirements and you may ignore.
", "rawMarkdown": "My apologies.\n\nThis was referencing a function that was previously available in Kaggle Notebooks/Kernels but we have since deprecated. We have updated the code requirements and you may ignore.", "votes": null }, { "id": "934949", "postDate": "07/19/2020 00:08:48", "content": "I understand about that. Thanks for your work 😃 \n
And another question is, can we use our own (not public) codes on Notebooks/Kernels by adding them as Private Dataset (like as locally trained models)?
", "rawMarkdown": "I understand about that. Thanks for your work 😃 \nAs long as the repo's license permits it, of course :)
", "rawMarkdown": "As long as the repo's license permits it, of course :)", "votes": null } ], "comments": [ { "id": 921143, "author_name": "snovik1975", "author_url": "", "post_date": "07/09/2020 05:18:22", "content": "I believe that since internet needs to be disabled in the submission notebook, it is not possible to install new packages. However I also believe that as kaggle staff sees common use of some package, they add it to the new versions of the docker
", "votes": null, "replies": [ { "id": 921208, "author_name": "shahules", "author_url": "", "post_date": "07/09/2020 06:04:09", "content": "It is possible to install packages without internet.
", "votes": null, "replies": [] } ] }, { "id": 921204, "author_name": "shahules", "author_url": "", "post_date": "07/09/2020 06:02:21", "content": "I think they meant some package which you made up locally and is not available to use publically.I am not sure though.
", "votes": null, "replies": [ { "id": 921978, "author_name": "ttahara", "author_url": "", "post_date": "07/09/2020 17:40:12", "content": "Thank you for reply. \nSo you mean that we can use package (or codes) such as open GitHub Repository which anyone can use ?
", "votes": null, "replies": [] }, { "id": 931148, "author_name": "shahules", "author_url": "", "post_date": "07/16/2020 02:45:37", "content": "yes,I think so.
", "votes": null, "replies": [] }, { "id": 977676, "author_name": "marcogorelli", "author_url": "", "post_date": "08/19/2020 16:30:47", "content": "As long as the repo's license permits it, of course :)
", "votes": null, "replies": [] } ] }, { "id": 921743, "author_name": "pranavkasela", "author_url": "", "post_date": "07/09/2020 14:10:50", "content": "Here is a similar question that was asked in the Toxic tweet competition: discussion
", "votes": null, "replies": [ { "id": 921951, "author_name": "ttahara", "author_url": "", "post_date": "07/09/2020 17:18:26", "content": "Thank you for information !!
\n\nI did not participate in the competition and missed the topic.
", "votes": null, "replies": [] } ] }, { "id": 930659, "author_name": "kneroma", "author_url": "", "post_date": "07/15/2020 16:17:55", "content": "This is a great question. Saying \"no custom packages enabled\" is meaningless in python, especially when \"custom datasets\" are enabled. A package is just a bunch of code files so what if I upload my own python files ? Are they custom packages ? Are they allowed ? I'm also confused !
", "votes": null, "replies": [ { "id": 930891, "author_name": "ttahara", "author_url": "", "post_date": "07/15/2020 19:38:06", "content": "Yes, I have the same question, too.\nIn the previous code competition, I uploaded my python code as a dataset and used it because the rule \"no custom packages enabled\" didn't exist.
", "votes": null, "replies": [] } ] }, { "id": 933497, "author_name": "mineshjethva", "author_url": "", "post_date": "07/17/2020 18:19:43", "content": "@addisonhoward can you please comment on this topic!
", "votes": null, "replies": [ { "id": 933546, "author_name": "addisonhoward", "author_url": "", "post_date": "07/17/2020 19:07:35", "content": "No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed. If you try to submit an output from a kernel with this enabled, you will be blocked on submission.
\n\nUsing code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.
", "votes": null, "replies": [] }, { "id": 933663, "author_name": "ttahara", "author_url": "", "post_date": "07/17/2020 21:52:38", "content": "@addisonhoward \nThank you for reply.\n
\n> No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed.
I think one source of confusion is that we cannot find “Packages” in kernel's \"Settings\". (Or you mean another \"Settings\"? )
\n\n
> Using code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.
\n\nI understand we can use public codes, pretrained model, external data by attaching them as dataset if and only if they are declared.
\nOn the other hand, in order to use our own (not public) code such as a custom model definition, we must code it on notebook or using utility script, right ?
\nCorrect me if I misunderstand. Thanks.
My apologies.
\n\nThis was referencing a function that was previously available in Kaggle Notebooks/Kernels but we have since deprecated. We have updated the code requirements and you may ignore.
", "votes": null, "replies": [] }, { "id": 934949, "author_name": "ttahara", "author_url": "", "post_date": "07/19/2020 00:08:48", "content": "I understand about that. Thanks for your work 😃 \n
And another question is, can we use our own (not public) codes on Notebooks/Kernels by adding them as Private Dataset (like as locally trained models)?
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "920421": "Hi.\nI'm confused about _custom packages_.\n\n[Code Requirements](https://www.kaggle.com/c/birdsong-recognition/overview/code-requirements) says _\"No custom packages enabled in kernels\"_. I think this means that we participants are **not allowed** to use custom packages in **_submission notebooks_**. \n\nI use some python packages not included in Kaggle Notebooks for **_training_** but don't use them in **_submission notebook_**. (If some parts are necessary, I code them.)\n\nSome public notebooks, however, **succeeded in submission** though those used codes or packages which are uploaded as Kaggle Dataset.\n\nMy question is simple, what _\"custom packages\"_ means ?\n\nThanks.", "921143": "I believe that since internet needs to be disabled in the submission notebook, it is not possible to install new packages. However I also believe that as kaggle staff sees common use of some package, they add it to the new versions of the docker", "921204": "I think they meant some package which you made up locally and is not available to use publically.I am not sure though.", "921208": "It is possible to install packages without internet.", "921743": "Here is a similar question that was asked in the Toxic tweet competition: [discussion](https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification/discussion/88021)", "921951": "Thank you for information !!\n\nI did not participate in the competition and missed the topic.", "921978": "Thank you for reply. \nSo you mean that we can use package (or codes) such as open GitHub Repository which anyone can use ?", "930659": "This is a great question. Saying \"no custom packages enabled\" is meaningless in python, especially when \"custom datasets\" are enabled. A package is just a bunch of code files so what if I upload my own python files ? Are they custom packages ? Are they allowed ? I'm also confused !", "930891": "Yes, I have the same question, too.\nIn the previous code competition, I uploaded my python code as a dataset and used it because the rule \"no custom packages enabled\" didn't exist.", "931148": "yes,I think so.", "933497": "addisonhoward can you please comment on this topic!", "933546": "No custom packages refers specifically to not being permitted to submit from a notebook where “Packages” under a kernel’s “Settings” are installed. If you try to submit an output from a kernel with this enabled, you will be blocked on submission.\n\nUsing code from GitHub as a pretrained model or external data is permitted, but must be declared on the external data thread.", "933663": "addisonhoward \nThank you for reply.\n