As this competition comes to an end, the majority are still having doubts on whether to submit compliant solutions or not as Kaggle's Team haven't given a clear answer to this problem. \nI suggest to automatically make all kernels public ( don't know if this is possible ) and with this participants can do the job and check compliant solutions and report non compliant ones.
", "messages": [ { "id": "956678", "postDate": "08/03/2020 17:26:20", "content": "As this competition comes to an end, the majority are still having doubts on whether to submit compliant solutions or not as Kaggle's Team haven't given a clear answer to this problem. \nI suggest to automatically make all kernels public ( don't know if this is possible ) and with this participants can do the job and check compliant solutions and report non compliant ones.
", "rawMarkdown": "As this competition comes to an end, the majority are still having doubts on whether to submit compliant solutions or not as Kaggle's Team haven't given a clear answer to this problem. \nI suggest to automatically make all kernels public ( don't know if this is possible ) and with this participants can do the job and check compliant solutions and report non compliant ones.", "votes": null }, { "id": "956886", "postDate": "08/03/2020 21:51:30", "content": "It's not possible. As you may notice, it's been 3 months that you haven't heard a clear response from the kaggle team or the organizers.
", "rawMarkdown": "It's not possible. As you may notice, it's been 3 months that you haven't heard a clear response from the kaggle team or the organizers.", "votes": null }, { "id": "956904", "postDate": "08/03/2020 22:26:11", "content": "Yeah but it's unfair to many great compliant solutions.. I'm sure they will find a solution to this.
", "rawMarkdown": "Yeah but it's unfair to many great compliant solutions.. I'm sure they will find a solution to this.", "votes": null }, { "id": "957281", "postDate": "08/04/2020 07:41:48", "content": "There is a clear answer. No team should use select any non-compliant solution as their final submissions. Please carefully read all pinned posts. They cannot be more clear on this.
", "rawMarkdown": "There is a clear answer. No team should use select any non-compliant solution as their final submissions. Please carefully read all pinned posts. They cannot be more clear on this.", "votes": null }, { "id": "957297", "postDate": "08/04/2020 07:52:16", "content": "@alexanderliao I expect the answer to be no teams can use non-compliant submissions as their final submissions. However I only get keywords like 'top teams' & 'winning teams'. This is still considered 'unclear' from my perspective. I do wish to get a clearer explanation than that. Also I've read every post, please don't make assumptions about my inadvertence, thank you. I actually took this very seriously and I want to abide the rules.
", "rawMarkdown": "alexanderliao I expect the answer to be **no teams can use non-compliant submissions as their final submissions**. However I only get keywords like 'top teams' & 'winning teams'. This is still considered 'unclear' from my perspective. I do wish to get a clearer explanation than that. Also I've read every post, please don't make assumptions about my inadvertence, thank you. I actually took this very seriously and I want to abide the rules.", "votes": null }, { "id": "957302", "postDate": "08/04/2020 07:57:43", "content": "I will just quote the words from the pinned post.
\n\nYou are free to try out any technique you like while crafting your submissions; however the two submissions that you select as your final submissions must be compliant with the competition rules.
For clarity the rules were not changed, MIT compatible license was required the entire time.
More restrictive open source licenses (e.g., GPL 2.0, GPL 3.0, others with similar restrictive characteristics) are not compatible with the MIT License and therefore should not be used in final submissions.
All of which are made regardless of one's leaderboard standings.
", "rawMarkdown": "I will just quote the words from the pinned post.\n\n`You are free to try out any technique you like while crafting your submissions; however the two submissions that you select as your final submissions must be compliant with the competition rules.`\n\n`For clarity the rules were not changed, MIT compatible license was required the entire time.`\n\n`More restrictive open source licenses (e.g., GPL 2.0, GPL 3.0, others with similar restrictive characteristics) are not compatible with the MIT License and therefore should not be used in final submissions.`\n\nAll of which are made regardless of one's leaderboard standings.", "votes": null }, { "id": "957312", "postDate": "08/04/2020 08:02:40", "content": "@alexanderliao I am fully aware of that, thanks. I saw ppl in the team merge sections saying they are using their own implementations of YOLO, therefore there will be no licensing issue. What is your opinion on that?
", "rawMarkdown": "alexanderliao I am fully aware of that, thanks. I saw ppl in the team merge sections saying they are using their own implementations of YOLO, therefore there will be no licensing issue. What is your opinion on that?", "votes": null }, { "id": "957317", "postDate": "08/04/2020 08:07:11", "content": "I will leave the decisions to the organizers, but personally, if you are referring to YoloV5 specifically, I don't think that it is possible to completely re-implement the entire codebase and reach a decent LB standing without looking at it or basing off it in a way and still manages to evade GPL. This dataset is extremely hard to optimize. I've tried over 20 models and only two are competitive LB-wise.
", "rawMarkdown": "I will leave the decisions to the organizers, but personally, if you are referring to YoloV5 specifically, I don't think that it is possible to completely re-implement the entire codebase and reach a decent LB standing without looking at it or basing off it in a way and still manages to evade GPL. This dataset is extremely hard to optimize. I've tried over 20 models and only two are competitive LB-wise.", "votes": null }, { "id": "957327", "postDate": "08/04/2020 08:14:59", "content": "I think what will end up happening is that you will see non-compliant teams gradually removed from the leaderboard in the upcoming weeks. Kaggle and the orgnizers tend to work hard to reproduce the winning results from competitions, which is a months-long process. I once won a competition and was working with the teams months after the private leaderboard is released. I trust that things will get sorted out then, so don't be discouraged by the temporary private LB results. I'm confident in Kaggle's determination in keeping all contest fair. For once, they banned a top-ranking grandmaster retroactively for cheating few months after the end of a competition.
", "rawMarkdown": "I think what will end up happening is that you will see non-compliant teams gradually removed from the leaderboard in the upcoming weeks. Kaggle and the orgnizers tend to work hard to reproduce the winning results from competitions, which is a months-long process. I once won a competition and was working with the teams months after the private leaderboard is released. I trust that things will get sorted out then, so don't be discouraged by the temporary private LB results. I'm confident in Kaggle's determination in keeping all contest fair. For once, they banned a top-ranking grandmaster retroactively for cheating few months after the end of a competition.", "votes": null } ], "comments": [ { "id": 956886, "author_name": "underwearfitting", "author_url": "", "post_date": "08/03/2020 21:51:30", "content": "It's not possible. As you may notice, it's been 3 months that you haven't heard a clear response from the kaggle team or the organizers.
", "votes": null, "replies": [ { "id": 956904, "author_name": "nacirbouazizi", "author_url": "", "post_date": "08/03/2020 22:26:11", "content": "Yeah but it's unfair to many great compliant solutions.. I'm sure they will find a solution to this.
", "votes": null, "replies": [] }, { "id": 957281, "author_name": "alexanderliao", "author_url": "", "post_date": "08/04/2020 07:41:48", "content": "There is a clear answer. No team should use select any non-compliant solution as their final submissions. Please carefully read all pinned posts. They cannot be more clear on this.
", "votes": null, "replies": [] }, { "id": 957297, "author_name": "underwearfitting", "author_url": "", "post_date": "08/04/2020 07:52:16", "content": "@alexanderliao I expect the answer to be no teams can use non-compliant submissions as their final submissions. However I only get keywords like 'top teams' & 'winning teams'. This is still considered 'unclear' from my perspective. I do wish to get a clearer explanation than that. Also I've read every post, please don't make assumptions about my inadvertence, thank you. I actually took this very seriously and I want to abide the rules.
", "votes": null, "replies": [] }, { "id": 957302, "author_name": "alexanderliao", "author_url": "", "post_date": "08/04/2020 07:57:43", "content": "I will just quote the words from the pinned post.
\n\nYou are free to try out any technique you like while crafting your submissions; however the two submissions that you select as your final submissions must be compliant with the competition rules.
For clarity the rules were not changed, MIT compatible license was required the entire time.
More restrictive open source licenses (e.g., GPL 2.0, GPL 3.0, others with similar restrictive characteristics) are not compatible with the MIT License and therefore should not be used in final submissions.
All of which are made regardless of one's leaderboard standings.
", "votes": null, "replies": [] }, { "id": 957312, "author_name": "underwearfitting", "author_url": "", "post_date": "08/04/2020 08:02:40", "content": "@alexanderliao I am fully aware of that, thanks. I saw ppl in the team merge sections saying they are using their own implementations of YOLO, therefore there will be no licensing issue. What is your opinion on that?
", "votes": null, "replies": [] }, { "id": 957317, "author_name": "alexanderliao", "author_url": "", "post_date": "08/04/2020 08:07:11", "content": "I will leave the decisions to the organizers, but personally, if you are referring to YoloV5 specifically, I don't think that it is possible to completely re-implement the entire codebase and reach a decent LB standing without looking at it or basing off it in a way and still manages to evade GPL. This dataset is extremely hard to optimize. I've tried over 20 models and only two are competitive LB-wise.
", "votes": null, "replies": [] } ] }, { "id": 957327, "author_name": "alexanderliao", "author_url": "", "post_date": "08/04/2020 08:14:59", "content": "I think what will end up happening is that you will see non-compliant teams gradually removed from the leaderboard in the upcoming weeks. Kaggle and the orgnizers tend to work hard to reproduce the winning results from competitions, which is a months-long process. I once won a competition and was working with the teams months after the private leaderboard is released. I trust that things will get sorted out then, so don't be discouraged by the temporary private LB results. I'm confident in Kaggle's determination in keeping all contest fair. For once, they banned a top-ranking grandmaster retroactively for cheating few months after the end of a competition.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "956678": "As this competition comes to an end, the majority are still having doubts on whether to submit compliant solutions or not as Kaggle's Team haven't given a clear answer to this problem. \nI suggest to automatically make all kernels public ( don't know if this is possible ) and with this participants can do the job and check compliant solutions and report non compliant ones.", "956886": "It's not possible. As you may notice, it's been 3 months that you haven't heard a clear response from the kaggle team or the organizers.", "956904": "Yeah but it's unfair to many great compliant solutions.. I'm sure they will find a solution to this.", "957281": "There is a clear answer. No team should use select any non-compliant solution as their final submissions. Please carefully read all pinned posts. They cannot be more clear on this.", "957297": "alexanderliao I expect the answer to be **no teams can use non-compliant submissions as their final submissions**. However I only get keywords like 'top teams' & 'winning teams'. This is still considered 'unclear' from my perspective. I do wish to get a clearer explanation than that. Also I've read every post, please don't make assumptions about my inadvertence, thank you. I actually took this very seriously and I want to abide the rules.", "957302": "I will just quote the words from the pinned post.\n\n`You are free to try out any technique you like while crafting your submissions; however the two submissions that you select as your final submissions must be compliant with the competition rules.`\n\n`For clarity the rules were not changed, MIT compatible license was required the entire time.`\n\n`More restrictive open source licenses (e.g., GPL 2.0, GPL 3.0, others with similar restrictive characteristics) are not compatible with the MIT License and therefore should not be used in final submissions.`\n\nAll of which are made regardless of one's leaderboard standings.", "957312": "alexanderliao I am fully aware of that, thanks. I saw ppl in the team merge sections saying they are using their own implementations of YOLO, therefore there will be no licensing issue. What is your opinion on that?", "957317": "I will leave the decisions to the organizers, but personally, if you are referring to YoloV5 specifically, I don't think that it is possible to completely re-implement the entire codebase and reach a decent LB standing without looking at it or basing off it in a way and still manages to evade GPL. This dataset is extremely hard to optimize. I've tried over 20 models and only two are competitive LB-wise.", "957327": "I think what will end up happening is that you will see non-compliant teams gradually removed from the leaderboard in the upcoming weeks. Kaggle and the orgnizers tend to work hard to reproduce the winning results from competitions, which is a months-long process. I once won a competition and was working with the teams months after the private leaderboard is released. I trust that things will get sorted out then, so don't be discouraged by the temporary private LB results. I'm confident in Kaggle's determination in keeping all contest fair. For once, they banned a top-ranking grandmaster retroactively for cheating few months after the end of a competition." }, "source": "meta" }