{ "id": 455939, "title": "How code requirements work - Submit Inference only or both training and Inference", "url": "/competitions/child-mind-institute-detect-sleep-states/discussion/455939", "author_name": "", "post_date": "2023-11-17T07:51:40.771659400Z", "votes": 2, "comment_count": 2, "views": 0, "content": "

I am a bit confused on how code competitions work. I have a couple of questions i need help on.
\nDoes the allocated time of 9hrs used for inference only or for both training and inference?
\nCan I do the data pre-processing and model training offline and import them to kaggle for inference only.
\nDo I need to create two different notebooks for training and Inference, how do I submit both of them, I assume kaggle is interested in both.
\nWhat is the difference between commit and submit in code competitions, I didn't fully understand the explanation in kaggle faqs
\nWhen the requirements state that internet is to be disabled, how to install packages
\nHow to recreate similar kaggle environment offline
\nIf possible share example-resources of such

", "messages": [ { "id": "2528217", "postDate": "11/17/2023 07:51:40", "content": "

I am a bit confused on how code competitions work. I have a couple of questions i need help on.
\nDoes the allocated time of 9hrs used for inference only or for both training and inference?
\nCan I do the data pre-processing and model training offline and import them to kaggle for inference only.
\nDo I need to create two different notebooks for training and Inference, how do I submit both of them, I assume kaggle is interested in both.
\nWhat is the difference between commit and submit in code competitions, I didn't fully understand the explanation in kaggle faqs
\nWhen the requirements state that internet is to be disabled, how to install packages
\nHow to recreate similar kaggle environment offline
\nIf possible share example-resources of such

", "rawMarkdown": "I am a bit confused on how code competitions work. I have a couple of questions i need help on.\nDoes the allocated time of 9hrs used for inference only or for both training and inference? \nCan I do the data pre-processing and model training offline and import them to kaggle for inference only. \nDo I need to create two different notebooks for training and Inference, how do I submit both of them, I assume kaggle is interested in both.\nWhat is the difference between commit and submit in code competitions, I didn't fully understand the explanation in kaggle faqs\nWhen the requirements state that internet is to be disabled, how to install packages\nHow to recreate similar kaggle environment offline\nIf possible share example-resources of such", "votes": null }, { "id": "2528331", "postDate": "11/17/2023 09:18:45", "content": "

You need to submit the model inference for sure (as this will be used to predict the hidden test set) @kenpachi99. You may also submit the training elements (this is not compulsory and is often counter-productive in such large datasets)

\n

Usually people train elsewhere (on Kaggle/ local PC) and save models to infer as a Kaggle dataset. This is then used to submit to the LB for evaluation.

", "rawMarkdown": "You need to submit the model inference for sure (as this will be used to predict the hidden test set) @kenpachi99. You may also submit the training elements (this is not compulsory and is often counter-productive in such large datasets)\n\nUsually people train elsewhere (on Kaggle/ local PC) and save models to infer as a Kaggle dataset. This is then used to submit to the LB for evaluation.", "votes": null }, { "id": "2528343", "postDate": "11/17/2023 09:38:14", "content": "

Thank you @ravi20076 , that explains alot

", "rawMarkdown": "Thank you @ravi20076 , that explains alot", "votes": null } ], "comments": [ { "id": 2528331, "author_name": "ravi20076", "author_url": "", "post_date": "11/17/2023 09:18:45", "content": "

You need to submit the model inference for sure (as this will be used to predict the hidden test set) @kenpachi99. You may also submit the training elements (this is not compulsory and is often counter-productive in such large datasets)

\n

Usually people train elsewhere (on Kaggle/ local PC) and save models to infer as a Kaggle dataset. This is then used to submit to the LB for evaluation.

", "votes": null, "replies": [ { "id": 2528343, "author_name": "kenpachi99", "author_url": "", "post_date": "11/17/2023 09:38:14", "content": "

Thank you @ravi20076 , that explains alot

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2528217": "I am a bit confused on how code competitions work. I have a couple of questions i need help on.\nDoes the allocated time of 9hrs used for inference only or for both training and inference? \nCan I do the data pre-processing and model training offline and import them to kaggle for inference only. \nDo I need to create two different notebooks for training and Inference, how do I submit both of them, I assume kaggle is interested in both.\nWhat is the difference between commit and submit in code competitions, I didn't fully understand the explanation in kaggle faqs\nWhen the requirements state that internet is to be disabled, how to install packages\nHow to recreate similar kaggle environment offline\nIf possible share example-resources of such", "2528331": "You need to submit the model inference for sure (as this will be used to predict the hidden test set) @kenpachi99. You may also submit the training elements (this is not compulsory and is often counter-productive in such large datasets)\n\nUsually people train elsewhere (on Kaggle/ local PC) and save models to infer as a Kaggle dataset. This is then used to submit to the LB for evaluation.", "2528343": "Thank you @ravi20076 , that explains alot" }, "source": "meta" }