![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1973217%2Fc57566a8a85a5d6db814a29e58344146%2FScreen%20Shot%202020-07-13%20at%2022.41.35.png?generation=1594647937120649&alt=media\")
\nSince it is near from merger deadline, I wish organizers fix this immediately.Now our team is try to team up with another team which has submission count 38.\nSo total submission should be 207 + 38 = 245, which is less than submission limit.\nBut kaggle is giving us the following error.\n\nSince it is near from merger deadline, I wish organizers fix this immediately.
Now our team is try to team up with another team which has submission count 38.\nSo total submission should be 207 + 38 = 245, which is less than submission limit.\nBut kaggle is giving us the following error.\n\nSince it is near from merger deadline, I wish organizers fix this immediately.
The count on the leaderboard doesn't include submissions which errored out. You can see the relevant total on the \"my submissions\" tab.
", "rawMarkdown": "The count on the leaderboard doesn't include submissions which errored out. You can see the relevant total on the \"my submissions\" tab. ", "votes": 3 }, { "id": 927691, "postDate": "2020-07-13T14:45:33.353Z", "content": "It seems that when merging, not only the displayed number of posts, but also the number of posts with an error are referenced. Does it match the total number including errors?
", "rawMarkdown": "It seems that when merging, not only the displayed number of posts, but also the number of posts with an error are referenced. Does it match the total number including errors?", "votes": 4, "replies": [ { "id": 927879, "postDate": "2020-07-13T16:02:12.477Z", "content": "Thanks!\nSo error submissions are not counted on LB, but are counted when merging teams?\nThis is a really terrible rule, I hope to get an answer from organizers on this matter.
", "rawMarkdown": "Thanks!\nSo error submissions are not counted on LB, but are counted when merging teams?\nThis is a really terrible rule, I hope to get an answer from organizers on this matter.", "votes": 3 }, { "id": 928270, "postDate": "2020-07-13T20:41:56.947Z", "content": "I imagine if your first few submissions all have errors, then you wouldn't have an lb score and thus on lb you essentially have 0 submissions. Therefore, it makes sense to not count submissions with errors on lb, although they could change it so that lb counts submissions with errors after the first one without I suppose
", "rawMarkdown": "I imagine if your first few submissions all have errors, then you wouldn't have an lb score and thus on lb you essentially have 0 submissions. Therefore, it makes sense to not count submissions with errors on lb, although they could change it so that lb counts submissions with errors after the first one without I suppose", "votes": 1 }, { "id": 928353, "postDate": "2020-07-13T22:59:40.450Z", "content": "Well I guess the issue here is that some people probe LB and waste submissions. So its hard to know if submission was not successful because some error in your code or because you probed the LB.
", "rawMarkdown": "Well I guess the issue here is that some people probe LB and waste submissions. So its hard to know if submission was not successful because some error in your code or because you probed the LB. ", "votes": 2 }, { "id": 928769, "postDate": "2020-07-14T07:50:12.923Z", "content": "Think with hidden test sets (or even not any like in Cornell Birds), it is pretty easy to get errors when trialling different architectures, etc. Some people just use the sample submission file if not the true Test submission to save GPU/TPU. While LB probing is bad, counting errors for team up seems not great either.
", "rawMarkdown": "Think with hidden test sets (or even not any like in Cornell Birds), it is pretty easy to get errors when trialling different architectures, etc. Some people just use the sample submission file if not the true Test submission to save GPU/TPU. While LB probing is bad, counting errors for team up seems not great either. ", "votes": 1 } ] }, { "id": 927890, "postDate": "2020-07-13T16:10:08.560Z", "content": "Thank you.
", "rawMarkdown": "Thank you.", "votes": 1 } ], "comments": [ { "id": 927876, "author_name": "Sohier Dane", "author_url": "", "post_date": "2020-07-13T16:01:00.320000", "content": "The count on the leaderboard doesn't include submissions which errored out. You can see the relevant total on the \"my submissions\" tab.
", "votes": 3, "replies": [] }, { "id": 927691, "author_name": "SiNpcw", "author_url": "", "post_date": "2020-07-13T14:45:33.353000", "content": "It seems that when merging, not only the displayed number of posts, but also the number of posts with an error are referenced. Does it match the total number including errors?
", "votes": 4, "replies": [ { "id": 927879, "author_name": "RabotniKuma", "author_url": "", "post_date": "2020-07-13T16:02:12.477000", "content": "Thanks!\nSo error submissions are not counted on LB, but are counted when merging teams?\nThis is a really terrible rule, I hope to get an answer from organizers on this matter.
", "votes": 3, "replies": [] }, { "id": 928270, "author_name": "Shujun", "author_url": "", "post_date": "2020-07-13T20:41:56.947000", "content": "I imagine if your first few submissions all have errors, then you wouldn't have an lb score and thus on lb you essentially have 0 submissions. Therefore, it makes sense to not count submissions with errors on lb, although they could change it so that lb counts submissions with errors after the first one without I suppose
", "votes": 1, "replies": [] }, { "id": 928353, "author_name": "DrHB", "author_url": "", "post_date": "2020-07-13T22:59:40.450000", "content": "Well I guess the issue here is that some people probe LB and waste submissions. So its hard to know if submission was not successful because some error in your code or because you probed the LB.
", "votes": 2, "replies": [] }, { "id": 928769, "author_name": "something4kag", "author_url": "", "post_date": "2020-07-14T07:50:12.923000", "content": "Think with hidden test sets (or even not any like in Cornell Birds), it is pretty easy to get errors when trialling different architectures, etc. Some people just use the sample submission file if not the true Test submission to save GPU/TPU. While LB probing is bad, counting errors for team up seems not great either.
", "votes": 1, "replies": [] } ] }, { "id": 927890, "author_name": "RabotniKuma", "author_url": "", "post_date": "2020-07-13T16:10:08.560000", "content": "Thank you.
", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "927578": "Now our team is try to team up with another team which has submission count 38.\nSo total submission should be 207 + 38 = 245, which is less than submission limit.\nBut kaggle is giving us the following error.\n\nSince it is near from merger deadline, I wish organizers fix this immediately.", "927876": "The count on the leaderboard doesn't include submissions which errored out. You can see the relevant total on the \"my submissions\" tab. ", "927691": "It seems that when merging, not only the displayed number of posts, but also the number of posts with an error are referenced. Does it match the total number including errors?", "927890": "Thank you." } }