Last 3 submissions of the notebooks are showing notebook timeout error, what should be done in this case?
", "messages": [ { "id": "1305083", "postDate": "05/13/2021 05:04:30", "content": "Last 3 submissions of the notebooks are showing notebook timeout error, what should be done in this case?
", "rawMarkdown": "Last 3 submissions of the notebooks are showing notebook timeout error, what should be done in this case?", "votes": null }, { "id": "1305592", "postDate": "05/13/2021 11:33:59", "content": "You should reduce a runtime of notebooks you submit.
", "rawMarkdown": "You should reduce a runtime of notebooks you submit.", "votes": null }, { "id": "1305638", "postDate": "05/13/2021 12:02:41", "content": "Okay, so what I am currently doing here is I am training the model for more number of epochs somewhere else, and then I am re-training the model again for few number of epochs in the kaggle submission notebook, I guess as each epoch is taking a little time, so may be while doing the submission this may be happening. So my idea is if I don't re-train it in kaggle submission kernel, and I do all the epochs training somewhere, will this help?
", "rawMarkdown": "Okay, so what I am currently doing here is I am training the model for more number of epochs somewhere else, and then I am re-training the model again for few number of epochs in the kaggle submission notebook, I guess as each epoch is taking a little time, so may be while doing the submission this may be happening. So my idea is if I don't re-train it in kaggle submission kernel, and I do all the epochs training somewhere, will this help?", "votes": null }, { "id": "1305664", "postDate": "05/13/2021 12:18:45", "content": "You shouldn't do any training in inference notebook, it takes too much time. Do all training outside the submission.
", "rawMarkdown": "You shouldn't do any training in inference notebook, it takes too much time. Do all training outside the submission.", "votes": null }, { "id": "1305669", "postDate": "05/13/2021 12:22:43", "content": "Yeah yeah, that's what I was saying. Anyways your suggestions are really good man, thank you for all the help that you did..😃😃.. I will keep asking questions from you..😁😁
", "rawMarkdown": "Yeah yeah, that's what I was saying. Anyways your suggestions are really good man, thank you for all the help that you did..😃😃.. I will keep asking questions from you..😁😁", "votes": null }, { "id": "1305677", "postDate": "05/13/2021 12:27:19", "content": "Also one thing, if I don't do any training in the inference notebook, then there is no need for GPU, right? Cause my GPU quota is about to finish.
", "rawMarkdown": "Also one thing, if I don't do any training in the inference notebook, then there is no need for GPU, right? Cause my GPU quota is about to finish.", "votes": null }, { "id": "1305808", "postDate": "05/13/2021 13:43:21", "content": "that depends, if you doing batch inference a GPU could help. Also I don't think the kernel evaluation for the leaderboard counts towards your GPU quota (I could be wrong though)
", "rawMarkdown": "that depends, if you doing batch inference a GPU could help. Also I don't think the kernel evaluation for the leaderboard counts towards your GPU quota (I could be wrong though)", "votes": null }, { "id": "1305965", "postDate": "05/13/2021 14:50:38", "content": "You should use GPU for inference, as it would be faster. As for quota, it will be counted only when you run your notebook. It won't be counted during the submission.
", "rawMarkdown": "You should use GPU for inference, as it would be faster. As for quota, it will be counted only when you run your notebook. It won't be counted during the submission.", "votes": null }, { "id": "1306054", "postDate": "05/13/2021 15:42:03", "content": "Okay… understood..
", "rawMarkdown": "Okay... understood..", "votes": null }, { "id": "1308856", "postDate": "05/15/2021 13:47:51", "content": "@atamazian tell me one thing, that we can make submission till 11 days, or only for 4 days we can make submission to this competition?
", "rawMarkdown": "atamazian tell me one thing, that we can make submission till 11 days, or only for 4 days we can make submission to this competition?", "votes": null } ], "comments": [ { "id": 1305592, "author_name": "atamazian", "author_url": "", "post_date": "05/13/2021 11:33:59", "content": "You should reduce a runtime of notebooks you submit.
", "votes": null, "replies": [ { "id": 1305638, "author_name": "jl18pg052", "author_url": "", "post_date": "05/13/2021 12:02:41", "content": "Okay, so what I am currently doing here is I am training the model for more number of epochs somewhere else, and then I am re-training the model again for few number of epochs in the kaggle submission notebook, I guess as each epoch is taking a little time, so may be while doing the submission this may be happening. So my idea is if I don't re-train it in kaggle submission kernel, and I do all the epochs training somewhere, will this help?
", "votes": null, "replies": [] }, { "id": 1305664, "author_name": "atamazian", "author_url": "", "post_date": "05/13/2021 12:18:45", "content": "You shouldn't do any training in inference notebook, it takes too much time. Do all training outside the submission.
", "votes": null, "replies": [] }, { "id": 1305669, "author_name": "jl18pg052", "author_url": "", "post_date": "05/13/2021 12:22:43", "content": "Yeah yeah, that's what I was saying. Anyways your suggestions are really good man, thank you for all the help that you did..😃😃.. I will keep asking questions from you..😁😁
", "votes": null, "replies": [] }, { "id": 1305677, "author_name": "jl18pg052", "author_url": "", "post_date": "05/13/2021 12:27:19", "content": "Also one thing, if I don't do any training in the inference notebook, then there is no need for GPU, right? Cause my GPU quota is about to finish.
", "votes": null, "replies": [] }, { "id": 1305808, "author_name": "coldfir3", "author_url": "", "post_date": "05/13/2021 13:43:21", "content": "that depends, if you doing batch inference a GPU could help. Also I don't think the kernel evaluation for the leaderboard counts towards your GPU quota (I could be wrong though)
", "votes": null, "replies": [] }, { "id": 1305965, "author_name": "atamazian", "author_url": "", "post_date": "05/13/2021 14:50:38", "content": "You should use GPU for inference, as it would be faster. As for quota, it will be counted only when you run your notebook. It won't be counted during the submission.
", "votes": null, "replies": [] }, { "id": 1306054, "author_name": "jl18pg052", "author_url": "", "post_date": "05/13/2021 15:42:03", "content": "Okay… understood..
", "votes": null, "replies": [] }, { "id": 1308856, "author_name": "jl18pg052", "author_url": "", "post_date": "05/15/2021 13:47:51", "content": "@atamazian tell me one thing, that we can make submission till 11 days, or only for 4 days we can make submission to this competition?
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1305083": "Last 3 submissions of the notebooks are showing notebook timeout error, what should be done in this case?", "1305592": "You should reduce a runtime of notebooks you submit.", "1305638": "Okay, so what I am currently doing here is I am training the model for more number of epochs somewhere else, and then I am re-training the model again for few number of epochs in the kaggle submission notebook, I guess as each epoch is taking a little time, so may be while doing the submission this may be happening. So my idea is if I don't re-train it in kaggle submission kernel, and I do all the epochs training somewhere, will this help?", "1305664": "You shouldn't do any training in inference notebook, it takes too much time. Do all training outside the submission.", "1305669": "Yeah yeah, that's what I was saying. Anyways your suggestions are really good man, thank you for all the help that you did..😃😃.. I will keep asking questions from you..😁😁", "1305677": "Also one thing, if I don't do any training in the inference notebook, then there is no need for GPU, right? Cause my GPU quota is about to finish.", "1305808": "that depends, if you doing batch inference a GPU could help. Also I don't think the kernel evaluation for the leaderboard counts towards your GPU quota (I could be wrong though)", "1305965": "You should use GPU for inference, as it would be faster. As for quota, it will be counted only when you run your notebook. It won't be counted during the submission.", "1306054": "Okay... understood..", "1308856": "atamazian tell me one thing, that we can make submission till 11 days, or only for 4 days we can make submission to this competition?" }, "source": "meta" }