![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8422589%2Fe5a4fb68b20397ceba16387e73fbdff4%2F2023-06-06%20%2018.30.09.png?generation=1686066381932735&alt=media\")
Hi all -
\nThe most recent LB update included an update to our 3.8 API. The public test set has not changed, but the change in APIs has solved an issue some people were experiencing previously, which may change public LB scores. In light of that, we are extending the competition by two weeks. The final results of the update will be published shortly.
", "messages": [ { "id": 2289156, "postDate": "2023-06-06T01:03:13.630Z", "content": "I terribly regret participating in this competition…
", "rawMarkdown": "I terribly regret participating in this competition…", "votes": 22 }, { "id": 2290324, "postDate": "2023-06-06T17:36:25.820Z", "content": "Hi everyone,
\nThank you all for expressing your concerns and frustrations! We know this has been a challenging competition, and we greatly appreciate the patience and perseverance of the community.
\nWe're working with Kaggle now on a post (will go up by today or tomorrow) that offers more explanation of the decisions made and the motivation behind the two week extension.
\nBest,
\nNatalie
Thanks for the clarification. It's a tough time for both the host and the community, and that's why communication is the most important thing in such a situation.
", "rawMarkdown": "Thanks for the clarification. It's a tough time for both the host and the community, and that's why communication is the most important thing in such a situation. ", "votes": 2 }, { "id": 2291226, "postDate": "2023-06-07T12:24:35.863Z", "content": "You should increase the prizes
", "rawMarkdown": "You should increase the prizes", "votes": 2 }, { "id": 2291556, "postDate": "2023-06-07T16:12:01.627Z", "content": "Thank you for your information, I am looking forward to the clarification. Could you please also provide the details of the common errors during the submissions? For example, \"Notebook Timeout\" means the inference time is longer than 9 hours. \"Notebook Threw Exception\" means there are some bugs in the code. The \"Submission Scoring Error\" is the most common bug I believe, but it is very vague. I am not sure if it is the problem during the scoring process (e.g., due to the wrong data type, or scoring functions from the Kaggle side is not robust enough, or it is something from our side, etc.). Thank you.
", "rawMarkdown": "Thank you for your information, I am looking forward to the clarification. Could you please also provide the details of the common errors during the submissions? For example, \"Notebook Timeout\" means the inference time is longer than 9 hours. \"Notebook Threw Exception\" means there are some bugs in the code. The \"Submission Scoring Error\" is the most common bug I believe, but it is very vague. I am not sure if it is the problem during the scoring process (e.g., due to the wrong data type, or scoring functions from the Kaggle side is not robust enough, or it is something from our side, etc.). Thank you.", "votes": 1 } ] }, { "id": 2290814, "postDate": "2023-06-07T06:01:18.960Z", "content": "The child has returned from studying and is participating in a Kaggle competition for the first time, and it seems like he has encountered the most outrageous situation in history. Please don't change anymore.🙏🏻🙏🏻🙏🏻
", "rawMarkdown": "The child has returned from studying and is participating in a Kaggle competition for the first time, and it seems like he has encountered the most outrageous situation in history. Please don't change anymore.🙏🏻🙏🏻🙏🏻", "votes": 14 }, { "id": 2288867, "postDate": "2023-06-05T18:17:44.227Z", "content": "Why… why though… At this point I just want this competition to end.
", "rawMarkdown": "Why... why though... At this point I just want this competition to end.", "votes": 12, "replies": [ { "id": 2288974, "postDate": "2023-06-05T20:33:49.467Z", "content": "Two weeks to find new one leak or error in API :)
", "rawMarkdown": "Two weeks to find new one leak or error in API :) ", "votes": 6, "replies": [ { "id": 2289030, "postDate": "2023-06-05T21:31:27.557Z", "content": "It wouldn't be a surprise.
", "rawMarkdown": "It wouldn't be a surprise.", "votes": 5 } ] } ] }, { "id": 2289293, "postDate": "2023-06-06T04:15:55.630Z", "content": "I suggest a preventive extension for 2 more years which should cover other cases of testing things in production
", "rawMarkdown": "I suggest a preventive extension for 2 more years which should cover other cases of testing things in production", "votes": 11 }, { "id": 2289231, "postDate": "2023-06-06T02:57:34.070Z", "content": "Please do not extend the deadline any further, as this would upset our strategy.
\nAlso, please answer the participant's question :
These are just examples, there are many more questions.
\nIf you had answered the questions sincerely and made a formal announcement, the participants would not have been so angry.
@philculliton
\nLet me ask you a question. You say above that \"The public test set has not changed\", but last time you said it would rerun with new test data. https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109
\nWill the test data eventually be updated and rerun? If Yes, What is the volume of test data and When will it be rerun?
I'm tired of unplanned extensions…
", "rawMarkdown": "I'm tired of unplanned extensions...", "votes": 4 }, { "id": 2289158, "postDate": "2023-06-06T01:05:20.877Z", "content": "My understanding is that the private test data will be updated, so I guess there is no contradiction to this part.
\nHowever, given there is no change to the public test set, why \"fixing the 3.8 API\" would lead to a change in \"public LB score\" is still a mystery.
\nMany fellows have pointed out that it might be related to the sorting (in my case, even sorting by index did not fully recover my original public scores). I wish we could have an official confirmation about this.
", "rawMarkdown": "My understanding is that the **private test data** will be updated, so I guess there is no contradiction to this part.\n\nHowever, given there is no change to the public test set, why \"fixing the 3.8 API\" would lead to a change in \"public LB score\" is still a mystery.\n\nMany fellows have pointed out that it might be related to the sorting (in my case, even sorting by index did not fully recover my original public scores). I wish we could have an official confirmation about this.", "votes": 1, "replies": [ { "id": 2289255, "postDate": "2023-06-06T03:28:46.393Z", "content": "The sorting part now is bit confusing ideally they should put a sample notebook on the prescribed submission way to avoid this confusion , or just pre-sort themselves like I guess before .
", "rawMarkdown": "The sorting part now is bit confusing ideally they should put a sample notebook on the prescribed submission way to avoid this confusion , or just pre-sort themselves like I guess before .", "votes": 2 } ] } ] }, { "id": 2289127, "postDate": "2023-06-05T23:54:04.200Z", "content": "How many extensions will satisfy you? I will be busy after June 15th, which will reduce the time I can allocate to the competition. This might even affect the color of the medal we might get.
", "rawMarkdown": "How many extensions will satisfy you? I will be busy after June 15th, which will reduce the time I can allocate to the competition. This might even affect the color of the medal we might get.", "votes": 7 }, { "id": 2289191, "postDate": "2023-06-06T02:10:38.063Z", "content": "Is the intern in charge of this comp? It is a terrible experience for us😭😭😭.
", "rawMarkdown": "Is the intern in charge of this comp? It is a terrible experience for us😭😭😭.", "votes": 6 }, { "id": 2290168, "postDate": "2023-06-06T15:46:30.510Z", "content": "Hi, @philculliton!
\nTell us what we should do with the fact that in some sessions (in 730 sessions) index does not increase sequentially by 1.
\nIf in the new api we sort test set by index, then for some sessions features based on elapsed_time will change. Because of this, some teams get a different LB score.
\nReferences
\n\n
I don't think they care, frankly, I don't even care anymore. I'd rather give up this <0.001 change if they will just end this competition on time….
", "rawMarkdown": "I don't think they care, frankly, I don't even care anymore. I'd rather give up this <0.001 change if they will just end this competition on time....", "votes": 1 } ] }, { "id": 2289792, "postDate": "2023-06-06T11:07:30.450Z", "content": "In fact, for participating team members, experiencing multiple competition delays is equivalent to being asked to increase their workload without reason. Even more terrifying than this is the official inability of Kaggle to test or promise the consistency of API scores. In the last 3.10 API update, many teams were forced to adjust their code, and time was wasted in meaningless places.
", "rawMarkdown": "In fact, for participating team members, experiencing multiple competition delays is equivalent to being asked to increase their workload without reason. Even more terrifying than this is the official inability of Kaggle to test or promise the consistency of API scores. In the last 3.10 API update, many teams were forced to adjust their code, and time was wasted in meaningless places.", "votes": 6 }, { "id": 2289629, "postDate": "2023-06-06T09:24:14.727Z", "content": "Hi @philculliton !
\nDespite everything, first I would like to thank you for your work. Things must have been very tough on your side as well.
However, can I ask is there an option to just keep everything as it currently is?
\nI understand some fellows might have experienced API issues since the last update, but I do believe many of us have already figured out some workarounds, keep improving our models, and finally arrived at this point.
\nI do not represent anyone but from the comments I could tell that many would prefer not to have any change again. Just let the competition end in the current state would be good enough.
", "rawMarkdown": "Hi @philculliton !\nDespite everything, first I would like to thank you for your work. Things must have been very tough on your side as well.\n\nHowever, can I ask is there an option to just **keep everything as it currently is**?\n\nI understand some fellows might have experienced API issues since the last update, but I do believe many of us have already figured out some workarounds, keep improving our models, and finally arrived at this point.\n\nI do not represent anyone but from the comments I could tell that many would prefer not to have any change again. Just let the competition end in the current state would be good enough.", "votes": 6, "replies": [ { "id": 2289728, "postDate": "2023-06-06T10:24:17.567Z", "content": "I believe this speaks for all of us.
", "rawMarkdown": "I believe this speaks for all of us. ", "votes": 2 } ] }, { "id": 2289229, "postDate": "2023-06-06T02:54:55.093Z", "content": "How about canceling the deadline?
", "rawMarkdown": "How about canceling the deadline?", "votes": 6 }, { "id": 2289189, "postDate": "2023-06-06T01:48:17.360Z", "content": "Why things always change? Most terrible competition I have ever been in.
", "rawMarkdown": "Why things always change? Most terrible competition I have ever been in.", "votes": 6 }, { "id": 2289133, "postDate": "2023-06-06T00:13:01.970Z", "content": "\nPlease answer participants' questions honestly and not one-sided information.
\nWe are confused.
My ranking suddenly went up dramatically? What happened?
", "rawMarkdown": "My ranking suddenly went up dramatically? What happened?", "votes": 3 }, { "id": 2289210, "postDate": "2023-06-06T02:37:58.337Z", "content": "Terrible..
", "rawMarkdown": "Terrible..", "votes": 3 }, { "id": 2289214, "postDate": "2023-06-06T02:44:33.747Z", "content": "oh man this has gone for ages now 😑
", "rawMarkdown": "oh man this has gone for ages now 😑 ", "votes": 4 }, { "id": 2290353, "postDate": "2023-06-06T18:04:27.890Z", "content": "Ohhhhh…..
", "rawMarkdown": "Ohhhhh.....", "votes": 1 }, { "id": 2290269, "postDate": "2023-06-06T16:55:01.090Z", "content": "My over 200 submissions just all got some random -0.07 score and now there is no way I learn any information from them since history public LB scores are now just some meaningless mess🫠
", "rawMarkdown": "My over 200 submissions just all got some random -0.07 score and now there is no way I learn any information from them since history public LB scores are now just some meaningless mess🫠", "votes": 1 }, { "id": 2290240, "postDate": "2023-06-06T16:37:12.487Z", "content": "Is the 3.10 api fixed?
", "rawMarkdown": "Is the 3.10 api fixed?", "votes": 1 }, { "id": 2290135, "postDate": "2023-06-06T15:21:15.657Z", "content": "Unbelievable competition, we have up to 3 LBs or more！
", "rawMarkdown": "Unbelievable competition, we have up to 3 LBs or more！", "votes": 1, "replies": [ { "id": 2290179, "postDate": "2023-06-06T15:55:10.210Z", "content": "Ye, very weird, it seems that quite a few top LB have been knocked down.
", "rawMarkdown": "Ye, very weird, it seems that quite a few top LB have been knocked down." } ] }, { "id": 2290116, "postDate": "2023-06-06T15:03:57.913Z", "content": "I was 53 in LB even three hours ago (have got 0.701 for nearly two months), but it change to 0.625 and 1300+ in LB now? a new error?
", "rawMarkdown": "I was 53 in LB even three hours ago (have got 0.701 for nearly two months), but it change to 0.625 and 1300+ in LB now? a new error?", "votes": 1, "replies": [ { "id": 2290117, "postDate": "2023-06-06T15:05:39.913Z", "content": "welcome to this gorgeous competition! 🎉
", "rawMarkdown": "welcome to this gorgeous competition! 🎉", "votes": 1 }, { "id": 2290260, "postDate": "2023-06-06T16:46:10.717Z", "content": "Same here. I got 0.703 with many notebooks but they died. I am desperately finding a solution…
", "rawMarkdown": "Same here. I got 0.703 with many notebooks but they died. I am desperately finding a solution...", "votes": 1, "replies": [ { "id": 2290278, "postDate": "2023-06-06T17:03:30.847Z", "content": "\n\nSame here. I got 0.703 with many notebooks but they died. I am desperately finding a solution…
\n
Try test.sort_by('index') in your submission, it won't give you the exactly same score but with some ±0.001 shake.
", "rawMarkdown": "> Same here. I got 0.703 with many notebooks but they died. I am desperately finding a solution...\n\nTry test.sort_by('index') in your submission, it won't give you the exactly same score but with some ±0.001 shake.\n", "votes": 1 }, { "id": 2290536, "postDate": "2023-06-06T23:17:56.917Z", "content": "Thank you for your will. I tried it, but still failed. It is weird that when I assigned the predictions by an array of 1s, it worked, but replace it by the prediction from my models, it failed.
", "rawMarkdown": "Thank you for your will. I tried it, but still failed. It is weird that when I assigned the predictions by an array of 1s, it worked, but replace it by the prediction from my models, it failed." }, { "id": 2290646, "postDate": "2023-06-07T02:26:40.787Z", "content": "Is there any difference in the 3.10 API? Can I use the old API for submission?
", "rawMarkdown": "Is there any difference in the 3.10 API? Can I use the old API for submission?" }, { "id": 2290659, "postDate": "2023-06-07T02:45:33.907Z", "content": "I am testing it, I will let you know if it works. Currently, I use this solution, https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/413004, but it still failed in my case. It is annoying that I have to use submissions to check how the API works. They should also call it \"hidden API\" as we don't know how it works.
", "rawMarkdown": "I am testing it, I will let you know if it works. Currently, I use this solution, https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/413004, but it still failed in my case. It is annoying that I have to use submissions to check how the API works. They should also call it \"hidden API\" as we don't know how it works." }, { "id": 2318764, "postDate": "2023-06-26T15:07:02.767Z", "content": "plz can you tell how did you solve it as i am facing the same issue, all my submissions are down, i was a little busy last month so haven't checked kaggle for a while and now i am surprised that the only submission that works is filling up with 1s.
", "rawMarkdown": "plz can you tell how did you solve it as i am facing the same issue, all my submissions are down, i was a little busy last month so haven't checked kaggle for a while and now i am surprised that the only submission that works is filling up with 1s." }, { "id": 2319794, "postDate": "2023-06-27T10:07:41.340Z", "content": "I detected that in the test data, there could be some new value in the categorical variables, then, encoding them returns NaN values.
", "rawMarkdown": "I detected that in the test data, there could be some new value in the categorical variables, then, encoding them returns NaN values." } ] } ] }, { "id": 2289289, "postDate": "2023-06-06T04:14:31.807Z", "content": "(this thread has got to be the most downvoted in the history of Kaggle)
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1842206%2F1ad477125e3b2cb1238a982635c57aab%2FBrojid.-Gang-up-against-you.jpg?generation=1686024962864379&alt=media\")
FYI: As far as I know it's this topic.
\nhttps://www.kaggle.com/competitions/aptos2019-blindness-detection/discussion/105510
HOLY!?!? hahaha, yes! You are correct!
", "rawMarkdown": "HOLY!?!? hahaha, yes! You are correct!" } ] } ] }, { "id": 2288829, "postDate": "2023-06-05T17:45:41.473Z", "content": "Hi all -
\nThe most recent LB update included an update to our 3.8 API. The public test set has not changed, but the change in APIs has solved an issue some people were experiencing previously, which may change public LB scores. In light of that, we are extending the competition by two weeks. The final results of the update will be published shortly.
", "rawMarkdown": "Hi all -\n\nThe most recent LB update included an update to our 3.8 API. The public test set has not changed, but the change in APIs has solved an issue some people were experiencing previously, which may change public LB scores. In light of that, we are extending the competition by two weeks. The final results of the update will be published shortly.", "votes": -79 }, { "id": 2289063, "postDate": "2023-06-05T22:18:20.193Z", "content": "Thanks, I got more time to work on this competition
", "rawMarkdown": "Thanks, I got more time to work on this competition", "votes": -15 }, { "id": 2289730, "postDate": "2023-06-06T10:25:18.477Z", "content": "Hi @philculliton !
\ndo not describe in cultural speech patterns what we as a whole community think about you:)
", "rawMarkdown": "Hi @philculliton !\n\ndo not describe in cultural speech patterns what we as a whole community think about you:)" }, { "id": 2303577, "postDate": "2023-06-15T11:09:01.610Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2290809, "postDate": "2023-06-07T05:59:37.147Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2290048, "postDate": "2023-06-06T14:09:17.017Z", "rawMarkdown": "", "votes": 1, "isDeleted": true }, { "id": 2289900, "postDate": "2023-06-06T12:25:29.130Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2288880, "postDate": "2023-06-05T18:28:07.990Z", "rawMarkdown": "", "votes": -5, "isDeleted": true } ], "comments": [ { "id": 2289156, "author_name": "Occccn", "author_url": "", "post_date": "2023-06-06T01:03:13.630000", "content": "I terribly regret participating in this competition…
", "votes": 22, "replies": [] }, { "id": 2290324, "author_name": "Natalie Rambis", "author_url": "", "post_date": "2023-06-06T17:36:25.820000", "content": "Hi everyone,
\nThank you all for expressing your concerns and frustrations! We know this has been a challenging competition, and we greatly appreciate the patience and perseverance of the community.
\nWe're working with Kaggle now on a post (will go up by today or tomorrow) that offers more explanation of the decisions made and the motivation behind the two week extension.
\nBest,
\nNatalie
Thanks for the clarification. It's a tough time for both the host and the community, and that's why communication is the most important thing in such a situation.
", "votes": 2, "replies": [] }, { "id": 2291226, "author_name": "Daniel Phalen", "author_url": "", "post_date": "2023-06-07T12:24:35.863000", "content": "You should increase the prizes
", "votes": 2, "replies": [] }, { "id": 2291556, "author_name": "Minh Tri Phan", "author_url": "", "post_date": "2023-06-07T16:12:01.627000", "content": "Thank you for your information, I am looking forward to the clarification. Could you please also provide the details of the common errors during the submissions? For example, \"Notebook Timeout\" means the inference time is longer than 9 hours. \"Notebook Threw Exception\" means there are some bugs in the code. The \"Submission Scoring Error\" is the most common bug I believe, but it is very vague. I am not sure if it is the problem during the scoring process (e.g., due to the wrong data type, or scoring functions from the Kaggle side is not robust enough, or it is something from our side, etc.). Thank you.
", "votes": 1, "replies": [] } ] }, { "id": 2290814, "author_name": "", "author_url": "", "post_date": "2023-06-07T06:01:18.960000", "content": "The child has returned from studying and is participating in a Kaggle competition for the first time, and it seems like he has encountered the most outrageous situation in history. Please don't change anymore.🙏🏻🙏🏻🙏🏻
", "votes": 14, "replies": [] }, { "id": 2288867, "author_name": "Woprime", "author_url": "", "post_date": "2023-06-05T18:17:44.227000", "content": "Why… why though… At this point I just want this competition to end.
", "votes": 12, "replies": [ { "id": 2288974, "author_name": "Remek Kinas", "author_url": "", "post_date": "2023-06-05T20:33:49.467000", "content": "Two weeks to find new one leak or error in API :)
", "votes": 6, "replies": [ { "id": 2289030, "author_name": "Woprime", "author_url": "", "post_date": "2023-06-05T21:31:27.557000", "content": "It wouldn't be a surprise.
", "votes": 5, "replies": [] } ] } ] }, { "id": 2289293, "author_name": "empty", "author_url": "", "post_date": "2023-06-06T04:15:55.630000", "content": "I suggest a preventive extension for 2 more years which should cover other cases of testing things in production
", "votes": 11, "replies": [] }, { "id": 2289231, "author_name": "shu421", "author_url": "", "post_date": "2023-06-06T02:57:34.070000", "content": "Please do not extend the deadline any further, as this would upset our strategy.
\nAlso, please answer the participant's question :
These are just examples, there are many more questions.
\nIf you had answered the questions sincerely and made a formal announcement, the participants would not have been so angry.
@philculliton
\nLet me ask you a question. You say above that \"The public test set has not changed\", but last time you said it would rerun with new test data. https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109
\nWill the test data eventually be updated and rerun? If Yes, What is the volume of test data and When will it be rerun?
I'm tired of unplanned extensions…
", "votes": 4, "replies": [] }, { "id": 2289158, "author_name": "Bosco Yung", "author_url": "", "post_date": "2023-06-06T01:05:20.877000", "content": "My understanding is that the private test data will be updated, so I guess there is no contradiction to this part.
\nHowever, given there is no change to the public test set, why \"fixing the 3.8 API\" would lead to a change in \"public LB score\" is still a mystery.
\nMany fellows have pointed out that it might be related to the sorting (in my case, even sorting by index did not fully recover my original public scores). I wish we could have an official confirmation about this.
", "votes": 1, "replies": [ { "id": 2289255, "author_name": "Gaurav Rawat", "author_url": "", "post_date": "2023-06-06T03:28:46.393000", "content": "The sorting part now is bit confusing ideally they should put a sample notebook on the prescribed submission way to avoid this confusion , or just pre-sort themselves like I guess before .
", "votes": 2, "replies": [] } ] } ] }, { "id": 2289127, "author_name": "Ryota", "author_url": "", "post_date": "2023-06-05T23:54:04.200000", "content": "How many extensions will satisfy you? I will be busy after June 15th, which will reduce the time I can allocate to the competition. This might even affect the color of the medal we might get.
", "votes": 7, "replies": [] }, { "id": 2289191, "author_name": "hyd", "author_url": "", "post_date": "2023-06-06T02:10:38.063000", "content": "Is the intern in charge of this comp? It is a terrible experience for us😭😭😭.
", "votes": 6, "replies": [] }, { "id": 2290168, "author_name": "superslon", "author_url": "", "post_date": "2023-06-06T15:46:30.510000", "content": "Hi, @philculliton!
\nTell us what we should do with the fact that in some sessions (in 730 sessions) index does not increase sequentially by 1.
\nIf in the new api we sort test set by index, then for some sessions features based on elapsed_time will change. Because of this, some teams get a different LB score.
\nReferences
\n\n
I don't think they care, frankly, I don't even care anymore. I'd rather give up this <0.001 change if they will just end this competition on time….
", "votes": 1, "replies": [] } ] }, { "id": 2289792, "author_name": "Dewei Chen", "author_url": "", "post_date": "2023-06-06T11:07:30.450000", "content": "In fact, for participating team members, experiencing multiple competition delays is equivalent to being asked to increase their workload without reason. Even more terrifying than this is the official inability of Kaggle to test or promise the consistency of API scores. In the last 3.10 API update, many teams were forced to adjust their code, and time was wasted in meaningless places.
", "votes": 6, "replies": [] }, { "id": 2289629, "author_name": "Bosco Yung", "author_url": "", "post_date": "2023-06-06T09:24:14.727000", "content": "Hi @philculliton !
\nDespite everything, first I would like to thank you for your work. Things must have been very tough on your side as well.
However, can I ask is there an option to just keep everything as it currently is?
\nI understand some fellows might have experienced API issues since the last update, but I do believe many of us have already figured out some workarounds, keep improving our models, and finally arrived at this point.
\nI do not represent anyone but from the comments I could tell that many would prefer not to have any change again. Just let the competition end in the current state would be good enough.
", "votes": 6, "replies": [ { "id": 2289728, "author_name": "Woprime", "author_url": "", "post_date": "2023-06-06T10:24:17.567000", "content": "I believe this speaks for all of us.
", "votes": 2, "replies": [] } ] }, { "id": 2289229, "author_name": "Sihan Lv", "author_url": "", "post_date": "2023-06-06T02:54:55.093000", "content": "How about canceling the deadline?
", "votes": 6, "replies": [] }, { "id": 2289189, "author_name": "Ethan", "author_url": "", "post_date": "2023-06-06T01:48:17.360000", "content": "Why things always change? Most terrible competition I have ever been in.
", "votes": 6, "replies": [] }, { "id": 2289133, "author_name": "Neko", "author_url": "", "post_date": "2023-06-06T00:13:01.970000", "content": "\nPlease answer participants' questions honestly and not one-sided information.
\nWe are confused.
My ranking suddenly went up dramatically? What happened?
", "votes": 3, "replies": [] }, { "id": 2289210, "author_name": "WU_DL", "author_url": "", "post_date": "2023-06-06T02:37:58.337000", "content": "Terrible..
", "votes": 3, "replies": [] }, { "id": 2289214, "author_name": "Gaurav Rawat", "author_url": "", "post_date": "2023-06-06T02:44:33.747000", "content": "oh man this has gone for ages now 😑
", "votes": 4, "replies": [] }, { "id": 2290353, "author_name": "Abdullah", "author_url": "", "post_date": "2023-06-06T18:04:27.890000", "content": "Ohhhhh…..
", "votes": 1, "replies": [] }, { "id": 2290269, "author_name": "Ya Xu", "author_url": "", "post_date": "2023-06-06T16:55:01.090000", "content": "My over 200 submissions just all got some random -0.07 score and now there is no way I learn any information from them since history public LB scores are now just some meaningless mess🫠
", "votes": 1, "replies": [] }, { "id": 2290240, "author_name": "pjmathematician", "author_url": "", "post_date": "2023-06-06T16:37:12.487000", "content": "Is the 3.10 api fixed?
", "votes": 1, "replies": [] }, { "id": 2290135, "author_name": "老肥", "author_url": "", "post_date": "2023-06-06T15:21:15.657000", "content": "Unbelievable competition, we have up to 3 LBs or more！
", "votes": 1, "replies": [ { "id": 2290179, "author_name": "Woprime", "author_url": "", "post_date": "2023-06-06T15:55:10.210000", "content": "Ye, very weird, it seems that quite a few top LB have been knocked down.
", "votes": 0, "replies": [] } ] }, { "id": 2290116, "author_name": "danniel", "author_url": "", "post_date": "2023-06-06T15:03:57.913000", "content": "I was 53 in LB even three hours ago (have got 0.701 for nearly two months), but it change to 0.625 and 1300+ in LB now? a new error?
", "votes": 1, "replies": [ { "id": 2290117, "author_name": "empty", "author_url": "", "post_date": "2023-06-06T15:05:39.913000", "content": "welcome to this gorgeous competition! 🎉
", "votes": 1, "replies": [] }, { "id": 2290260, "author_name": "Minh Tri Phan", "author_url": "", "post_date": "2023-06-06T16:46:10.717000", "content": "Same here. I got 0.703 with many notebooks but they died. I am desperately finding a solution…
", "votes": 1, "replies": [ { "id": 2290278, "author_name": "Ya Xu", "author_url": "", "post_date": "2023-06-06T17:03:30.847000", "content": "\n\nSame here. I got 0.703 with many notebooks but they died. I am desperately finding a solution…
\n
Try test.sort_by('index') in your submission, it won't give you the exactly same score but with some ±0.001 shake.
", "votes": 1, "replies": [] }, { "id": 2290536, "author_name": "Minh Tri Phan", "author_url": "", "post_date": "2023-06-06T23:17:56.917000", "content": "Thank you for your will. I tried it, but still failed. It is weird that when I assigned the predictions by an array of 1s, it worked, but replace it by the prediction from my models, it failed.
", "votes": 0, "replies": [] }, { "id": 2290646, "author_name": "biubiubiu~", "author_url": "", "post_date": "2023-06-07T02:26:40.787000", "content": "Is there any difference in the 3.10 API? Can I use the old API for submission?
", "votes": 0, "replies": [] }, { "id": 2290659, "author_name": "Minh Tri Phan", "author_url": "", "post_date": "2023-06-07T02:45:33.907000", "content": "I am testing it, I will let you know if it works. Currently, I use this solution, https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/413004, but it still failed in my case. It is annoying that I have to use submissions to check how the API works. They should also call it \"hidden API\" as we don't know how it works.
", "votes": 0, "replies": [] }, { "id": 2318764, "author_name": "kaoutar", "author_url": "", "post_date": "2023-06-26T15:07:02.767000", "content": "plz can you tell how did you solve it as i am facing the same issue, all my submissions are down, i was a little busy last month so haven't checked kaggle for a while and now i am surprised that the only submission that works is filling up with 1s.
", "votes": 0, "replies": [] }, { "id": 2319794, "author_name": "Minh Tri Phan", "author_url": "", "post_date": "2023-06-27T10:07:41.340000", "content": "I detected that in the test data, there could be some new value in the categorical variables, then, encoding them returns NaN values.
", "votes": 0, "replies": [] } ] } ] }, { "id": 2289289, "author_name": "BwandoWando", "author_url": "", "post_date": "2023-06-06T04:14:31.807000", "content": "(this thread has got to be the most downvoted in the history of Kaggle)
\n
FYI: As far as I know it's this topic.
\nhttps://www.kaggle.com/competitions/aptos2019-blindness-detection/discussion/105510
HOLY!?!? hahaha, yes! You are correct!
", "votes": 0, "replies": [] } ] } ] }, { "id": 2289063, "author_name": "william.wu", "author_url": "", "post_date": "2023-06-05T22:18:20.193000", "content": "Thanks, I got more time to work on this competition
", "votes": -15, "replies": [] }, { "id": 2289730, "author_name": "Pavel Pavlov", "author_url": "", "post_date": "2023-06-06T10:25:18.477000", "content": "Hi @philculliton !
\ndo not describe in cultural speech patterns what we as a whole community think about you:)
", "votes": 0, "replies": [] }, { "id": 2303577, "author_name": "", "author_url": "", "post_date": "2023-06-15T11:09:01.610000", "content": "", "votes": 0, "replies": [] }, { "id": 2290809, "author_name": "", "author_url": "", "post_date": "2023-06-07T05:59:37.147000", "content": "", "votes": 0, "replies": [] }, { "id": 2290048, "author_name": "", "author_url": "", "post_date": "2023-06-06T14:09:17.017000", "content": "", "votes": 1, "replies": [] }, { "id": 2289900, "author_name": "", "author_url": "", "post_date": "2023-06-06T12:25:29.130000", "content": "", "votes": 0, "replies": [] }, { "id": 2288880, "author_name": "", "author_url": "", "post_date": "2023-06-05T18:28:07.990000", "content": "", "votes": -5, "replies": [] } ], "raw_markdown_by_id": { "2289156": "I terribly regret participating in this competition…", "2290324": "Hi everyone, \n\nThank you all for expressing your concerns and frustrations! We know this has been a challenging competition, and we greatly appreciate the patience and perseverance of the community. \n\nWe're working with Kaggle now on a post (will go up by today or tomorrow) that offers more explanation of the decisions made and the motivation behind the two week extension. \n\nBest,\nNatalie", "2290814": "The child has returned from studying and is participating in a Kaggle competition for the first time, and it seems like he has encountered the most outrageous situation in history. Please don't change anymore.🙏🏻🙏🏻🙏🏻", "2288867": "Why... why though... At this point I just want this competition to end.", "2289293": "I suggest a preventive extension for 2 more years which should cover other cases of testing things in production", "2289231": "Please do not extend the deadline any further, as this would upset our strategy.\nAlso, please answer the participant's question :\n\n- https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109#2284890\n- https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109#2286587\n\nThese are just examples, there are many more questions.\nIf you had answered the questions sincerely and made a formal announcement, the participants would not have been so angry.", "2288956": "@philculliton \nLet me ask you a question. You say above that \"The public test set has not changed\", but last time you said it would rerun with **new test data**. [https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109](https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109)\nWill the test data eventually be updated and rerun? If Yes, What is the volume of test data and When will it be rerun?", "2289127": "How many extensions will satisfy you? I will be busy after June 15th, which will reduce the time I can allocate to the competition. This might even affect the color of the medal we might get.", "2289191": "Is the intern in charge of this comp? It is a terrible experience for us😭😭😭.", "2290168": "Hi, @philculliton!\n\nTell us what we should do with the fact that in some sessions (in 730 sessions) index does not increase sequentially by 1. \n\nIf in the new api we sort test set by index, then for some sessions features based on elapsed_time will change. Because of this, some teams get a different LB score. \n\nReferences\n- [great EDA made by @abaojiang ](https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/395250)\n- [comment in other thread made by @chaudharypriyanshu ](https://www.kaggle.com/competitions/predict-student-performance-from-game-play/discussion/414109#2286103)\n\n", "2289792": "In fact, for participating team members, experiencing multiple competition delays is equivalent to being asked to increase their workload without reason. Even more terrifying than this is the official inability of Kaggle to test or promise the consistency of API scores. In the last 3.10 API update, many teams were forced to adjust their code, and time was wasted in meaningless places.", "2289629": "Hi @philculliton !\nDespite everything, first I would like to thank you for your work. Things must have been very tough on your side as well.\n\nHowever, can I ask is there an option to just **keep everything as it currently is**?\n\nI understand some fellows might have experienced API issues since the last update, but I do believe many of us have already figured out some workarounds, keep improving our models, and finally arrived at this point.\n\nI do not represent anyone but from the comments I could tell that many would prefer not to have any change again. Just let the competition end in the current state would be good enough.", "2289229": "How about canceling the deadline?", "2289189": "Why things always change? Most terrible competition I have ever been in.", "2289133": "@philculliton\n\nPlease answer participants' questions honestly and not one-sided information.\nWe are confused.", "2291116": "My ranking suddenly went up dramatically? What happened?", "2289210": "Terrible..", "2289214": "oh man this has gone for ages now 😑 ", "2290353": "Ohhhhh.....", "2290269": "My over 200 submissions just all got some random -0.07 score and now there is no way I learn any information from them since history public LB scores are now just some meaningless mess🫠", "2290240": "Is the 3.10 api fixed?", "2290135": "Unbelievable competition, we have up to 3 LBs or more！", "2290116": "I was 53 in LB even three hours ago (have got 0.701 for nearly two months), but it change to 0.625 and 1300+ in LB now? a new error?", "2289289": "(this thread has got to be the most downvoted in the history of Kaggle)\n\n", "2288829": "Hi all -\n\nThe most recent LB update included an update to our 3.8 API. The public test set has not changed, but the change in APIs has solved an issue some people were experiencing previously, which may change public LB scores. In light of that, we are extending the competition by two weeks. The final results of the update will be published shortly.", "2289063": "Thanks, I got more time to work on this competition", "2289730": "Hi @philculliton !\n\ndo not describe in cultural speech patterns what we as a whole community think about you:)", "2303577": "", "2290809": "", "2290048": "", "2289900": "", "2288880": "" } }