Officially, the running time of the submit code should meet the following requirements: GPU<=6h.\nBut I found that the submit code which runs for 8 hours can still get a score without a timeout error. \nSo, which is the following statement is right? or other right explanations？\n1. Getting scores when submitting means meeting the requirements. So, as long as you get the score, you don’t need to care about the running time (even exceeds 6h).\n2. Although 8h codes can get scores now, such codes will be disqualified due to time limitations in the final evaluation. Therefore, when submitting, it is necessary to strictly control the running time to be less than 6h.
\n\nThe above is the situation when using GPU, but using CPU is the same.
\n\nHope someone can help me！
", "messages": [ { "id": "954631", "postDate": "08/01/2020 23:46:32", "content": "Officially, the running time of the submit code should meet the following requirements: GPU<=6h.\nBut I found that the submit code which runs for 8 hours can still get a score without a timeout error. \nSo, which is the following statement is right? or other right explanations？\n1. Getting scores when submitting means meeting the requirements. So, as long as you get the score, you don’t need to care about the running time (even exceeds 6h).\n2. Although 8h codes can get scores now, such codes will be disqualified due to time limitations in the final evaluation. Therefore, when submitting, it is necessary to strictly control the running time to be less than 6h.
\n\nThe above is the situation when using GPU, but using CPU is the same.
\n\nHope someone can help me！
", "rawMarkdown": "Officially, the running time of the submit code should meet the following requirements: GPU<=6h.\nBut I found that the submit code which runs for 8 hours can still get a score without a timeout error. \nSo, which is the following statement is right? or other right explanations？\n1. Getting scores when submitting means meeting the requirements. So, as long as you get the score, you don’t need to care about the running time (even exceeds 6h).\n2. Although 8h codes can get scores now, such codes will be disqualified due to time limitations in the final evaluation. Therefore, when submitting, it is necessary to strictly control the running time to be less than 6h.\n\nThe above is the situation when using GPU, but using CPU is the same.\n\nHope someone can help me！", "votes": null }, { "id": "954720", "postDate": "08/02/2020 02:54:04", "content": "I have the same concerns as you.
", "rawMarkdown": "I have the same concerns as you.", "votes": null }, { "id": "955211", "postDate": "08/02/2020 12:57:24", "content": "The test data used for private LB is less than that used for public LB. So don't worry about the submission time!
", "rawMarkdown": "The test data used for private LB is less than that used for public LB. So don't worry about the submission time!", "votes": null }, { "id": "955229", "postDate": "08/02/2020 13:23:45", "content": "I thought both public and private data is used when we submit our solution?
", "rawMarkdown": "I thought both public and private data is used when we submit our solution?", "votes": null }, { "id": "955245", "postDate": "08/02/2020 13:35:09", "content": "I've heard many versions of explanations and guesses. The official one is undoubtedly the most reliable, so what should we do to get an official reply?
", "rawMarkdown": "I've heard many versions of explanations and guesses. The official one is undoubtedly the most reliable, so what should we do to get an official reply?", "votes": null }, { "id": "955275", "postDate": "08/02/2020 14:07:37", "content": "The first one seems right. once your submission gets a score, it's good to go. it's already finished scoring on test set.\nalso check out this post. \nhope it will clear the doubts.
", "rawMarkdown": "The first one seems right. once your submission gets a score, it's good to go. it's already finished scoring on test set.\nalso check out this[ post. ](https://www.kaggle.com/c/global-wheat-detection/discussion/168944)\nhope it will clear the doubts.", "votes": null }, { "id": "955301", "postDate": "08/02/2020 14:35:25", "content": "It helps me a lot! Thanks so much!
", "rawMarkdown": "It helps me a lot! Thanks so much!", "votes": null }, { "id": "955421", "postDate": "08/02/2020 16:14:56", "content": "The organizers have confirmed that when you run your submission it is run on both the private and public test data, not just the public 62% (Source).
\n\nBut I agree, the timing issue is concerning. The code requirements states:
\n\n\n\n\nIn order for the \"Submit to Competition\" button to be active after a commit, the following conditions must be met:\n - GPU Notebook <= 6 hours run-time
\n
Which is true. However, when we actually submit it the timing appears to allow submissions up to 9 hours. This is where the concern arises. I would love to see Kaggle ( @maggiemd @inversion @philculliton ) or the organizers ( @bendvd @stavness ) address this quickly before the end of the competition.
", "rawMarkdown": "The organizers have confirmed that when you run your submission it is run on both the private and public test data, not just the public 62% ([Source](https://www.kaggle.com/c/global-wheat-detection/discussion/161081#900989)).\n\nBut I agree, the timing issue is concerning. The code requirements states:\n> **In order for the \"Submit to Competition\" button to be active after a commit**, the following conditions must be met:\n - GPU Notebook <= 6 hours run-time\n\nWhich is true. However, when we actually submit it the timing appears to allow submissions up to 9 hours. This is where the concern arises. I would love to see Kaggle ( @maggiemd @inversion @philculliton ) or the organizers ( @bendvd @stavness ) address this quickly before the end of the competition.", "votes": null }, { "id": "957413", "postDate": "08/04/2020 09:30:43", "content": "I can't submit when it cost more than 6h.
", "rawMarkdown": "I can't submit when it cost more than 6h.", "votes": null }, { "id": "957780", "postDate": "08/04/2020 14:56:44", "content": "Right. But when you submit on the full set it can take longer. That is where the confusion arises.
", "rawMarkdown": "Right. But when you submit on the full set it can take longer. That is where the confusion arises.", "votes": null } ], "comments": [ { "id": 954720, "author_name": "nvnnghia", "author_url": "", "post_date": "08/02/2020 02:54:04", "content": "I have the same concerns as you.
", "votes": null, "replies": [ { "id": 955211, "author_name": "zacchaeus", "author_url": "", "post_date": "08/02/2020 12:57:24", "content": "The test data used for private LB is less than that used for public LB. So don't worry about the submission time!
", "votes": null, "replies": [] }, { "id": 955229, "author_name": "nvnnghia", "author_url": "", "post_date": "08/02/2020 13:23:45", "content": "I thought both public and private data is used when we submit our solution?
", "votes": null, "replies": [] }, { "id": 955245, "author_name": "stonewst98", "author_url": "", "post_date": "08/02/2020 13:35:09", "content": "I've heard many versions of explanations and guesses. The official one is undoubtedly the most reliable, so what should we do to get an official reply?
", "votes": null, "replies": [] }, { "id": 955421, "author_name": "matthewmasters", "author_url": "", "post_date": "08/02/2020 16:14:56", "content": "The organizers have confirmed that when you run your submission it is run on both the private and public test data, not just the public 62% (Source).
\n\nBut I agree, the timing issue is concerning. The code requirements states:
\n\n\n\n\nIn order for the \"Submit to Competition\" button to be active after a commit, the following conditions must be met:\n - GPU Notebook <= 6 hours run-time
\n
Which is true. However, when we actually submit it the timing appears to allow submissions up to 9 hours. This is where the concern arises. I would love to see Kaggle ( @maggiemd @inversion @philculliton ) or the organizers ( @bendvd @stavness ) address this quickly before the end of the competition.
", "votes": null, "replies": [] } ] }, { "id": 955275, "author_name": "cooleel", "author_url": "", "post_date": "08/02/2020 14:07:37", "content": "The first one seems right. once your submission gets a score, it's good to go. it's already finished scoring on test set.\nalso check out this post. \nhope it will clear the doubts.
", "votes": null, "replies": [ { "id": 955301, "author_name": "stonewst98", "author_url": "", "post_date": "08/02/2020 14:35:25", "content": "It helps me a lot! Thanks so much!
", "votes": null, "replies": [] } ] }, { "id": 957413, "author_name": "zgh19971022", "author_url": "", "post_date": "08/04/2020 09:30:43", "content": "I can't submit when it cost more than 6h.
", "votes": null, "replies": [ { "id": 957780, "author_name": "matthewmasters", "author_url": "", "post_date": "08/04/2020 14:56:44", "content": "Right. But when you submit on the full set it can take longer. That is where the confusion arises.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "954631": "Officially, the running time of the submit code should meet the following requirements: GPU<=6h.\nBut I found that the submit code which runs for 8 hours can still get a score without a timeout error. \nSo, which is the following statement is right? or other right explanations？\n1. Getting scores when submitting means meeting the requirements. So, as long as you get the score, you don’t need to care about the running time (even exceeds 6h).\n2. Although 8h codes can get scores now, such codes will be disqualified due to time limitations in the final evaluation. Therefore, when submitting, it is necessary to strictly control the running time to be less than 6h.\n\nThe above is the situation when using GPU, but using CPU is the same.\n\nHope someone can help me！", "954720": "I have the same concerns as you.", "955211": "The test data used for private LB is less than that used for public LB. So don't worry about the submission time!", "955229": "I thought both public and private data is used when we submit our solution?", "955245": "I've heard many versions of explanations and guesses. The official one is undoubtedly the most reliable, so what should we do to get an official reply?", "955275": "The first one seems right. once your submission gets a score, it's good to go. it's already finished scoring on test set.\nalso check out this[ post. ](https://www.kaggle.com/c/global-wheat-detection/discussion/168944)\nhope it will clear the doubts.", "955301": "It helps me a lot! Thanks so much!", "955421": "The organizers have confirmed that when you run your submission it is run on both the private and public test data, not just the public 62% ([Source](https://www.kaggle.com/c/global-wheat-detection/discussion/161081#900989)).\n\nBut I agree, the timing issue is concerning. The code requirements states:\n> **In order for the \"Submit to Competition\" button to be active after a commit**, the following conditions must be met:\n - GPU Notebook <= 6 hours run-time\n\nWhich is true. However, when we actually submit it the timing appears to allow submissions up to 9 hours. This is where the concern arises. I would love to see Kaggle ( @maggiemd @inversion @philculliton ) or the organizers ( @bendvd @stavness ) address this quickly before the end of the competition.", "957413": "I can't submit when it cost more than 6h.", "957780": "Right. But when you submit on the full set it can take longer. That is where the confusion arises." }, "source": "meta" }