As you might know, a wonderful event was held this weekend - Kaggledays Paris. I was doing a workshop there called \"Dirty tricks for image classification\" (video should be available soon).
\n\nAmong several tricks I shared with the participants was googling file names from the test set. Guess what? It also works for text classification :)
\n\nTry to google 00c660c58101eafe25e9. There is going to be only one result: this link https://pastebin.com/KeVgnB6L which looks very much like a submission. To save your attempts, it scores 0.699 on LB. It was uploaded on NOV 16TH, 2018 - the exact same date when someone called lcch got the same result (see raw data).
\n\nAnother day, another \"leak\"... another valuable lesson for @lcch135616 ;)
", "messages": [ { "id": 462981, "postDate": "2019-01-29T07:26:21.717Z", "content": "As you might know, a wonderful event was held this weekend - Kaggledays Paris. I was doing a workshop there called \"Dirty tricks for image classification\" (video should be available soon).
\n\nAmong several tricks I shared with the participants was googling file names from the test set. Guess what? It also works for text classification :)
\n\nTry to google 00c660c58101eafe25e9. There is going to be only one result: this link https://pastebin.com/KeVgnB6L which looks very much like a submission. To save your attempts, it scores 0.699 on LB. It was uploaded on NOV 16TH, 2018 - the exact same date when someone called lcch got the same result (see raw data).
\n\nAnother day, another \"leak\"... another valuable lesson for @lcch135616 ;)
", "rawMarkdown": "As you might know, a wonderful event was held this weekend - Kaggledays Paris. I was doing a workshop there called \"Dirty tricks for image classification\" (video should be available soon).\n\nAmong several tricks I shared with the participants was googling file names from the test set. Guess what? It also works for text classification :)\n\nTry to google *00c660c58101eafe25e9*. There is going to be only one result: this link https://pastebin.com/KeVgnB6L which looks very much like a submission. To save your attempts, it scores 0.699 on LB. It was uploaded on NOV 16TH, 2018 - the exact same date when someone called **lcch** got the same result (see [raw data][1]).\n\nAnother day, another \"leak\"... another valuable lesson for @lcch135616 ;)\n\n [1]: https://www.kaggle.com/c/10737/publicleaderboarddata.zip", "votes": 56 }, { "id": 463158, "postDate": "2019-01-29T14:48:42.567Z", "content": "I mark this trick as Insincere :)
", "rawMarkdown": "I mark this trick as **Insincere** :) ", "votes": 10, "replies": [ { "id": 463171, "postDate": "2019-01-29T15:12:54.487Z", "content": "On the second thoughts, I was wondering how a submission ended up there. This potentially means offline teaming or solution sharing of some kind. Puts a real question in my head;
\n\nthere are no reasons to trust public LB at this competition, regardless of this submission
", "rawMarkdown": "there are no reasons to trust public LB at this competition, regardless of this submission", "votes": 8 }, { "id": 463203, "postDate": "2019-01-29T15:56:44.177Z", "content": "True. I agree
", "rawMarkdown": "True. I agree", "votes": -3 }, { "id": 463963, "postDate": "2019-01-31T00:48:58.337Z", "content": "for newcomers like me, LB just gets you anxious! Paulo keeps reminding me to forget LB... Easier said than done...
", "rawMarkdown": "for newcomers like me, LB just gets you anxious! Paulo keeps reminding me to forget LB... Easier said than done...", "votes": 1 } ] }, { "id": 463361, "postDate": "2019-01-29T21:29:42.013Z", "content": "I don't know what you all are upvoting here, folks. Would you like such dirty hacks to be well-known and have many more dummy kagglers on the leaderboard?
", "rawMarkdown": "I don't know what you all are upvoting here, folks. Would you like such dirty hacks to be well-known and have many more dummy kagglers on the leaderboard? ", "votes": 10, "replies": [ { "id": 463561, "postDate": "2019-01-30T07:28:06.357Z", "content": "the post actually was about not leaving your private data on https://pastebin.com
", "rawMarkdown": "the post actually was about not leaving your private data on https://pastebin.com", "votes": 5 }, { "id": 463581, "postDate": "2019-01-30T08:22:02.853Z", "content": "Of course I got it. But at the same time for many others the post can serve as a call to cheating.
", "rawMarkdown": "Of course I got it. But at the same time for many others the post can serve as a call to cheating.", "votes": 2 }, { "id": 464659, "postDate": "2019-02-01T08:41:33.633Z", "content": "Yury we should appreciate everyone
", "rawMarkdown": "Yury we should appreciate everyone", "votes": 1 } ] }, { "id": 462990, "postDate": "2019-01-29T07:46:17.523Z", "content": "Really a very Dirty Trick :)
", "rawMarkdown": "Really a very Dirty Trick :)", "votes": 3 }, { "id": 463163, "postDate": "2019-01-29T14:54:46.043Z", "content": "True detective :D
", "rawMarkdown": "True detective :D", "votes": 4, "replies": [ { "id": 463196, "postDate": "2019-01-29T15:41:23.990Z", "content": "that what makes Kaggle so fun for me, what gives me drive
", "rawMarkdown": "that what makes Kaggle so fun for me, what gives me drive", "votes": 4 } ] }, { "id": 463098, "postDate": "2019-01-29T11:51:51.287Z", "content": "no use
", "rawMarkdown": "no use", "votes": 3 }, { "id": 464657, "postDate": "2019-02-01T08:38:51.137Z", "content": "Thnx We know
", "rawMarkdown": "Thnx We know", "votes": 1 }, { "id": 463474, "postDate": "2019-01-30T03:33:02.783Z", "content": "Probably this is useful to check where our models predict wrong and fix the model a little bit ....
", "rawMarkdown": "Probably this is useful to check where our models predict wrong and fix the model a little bit ....", "votes": 1 }, { "id": 463454, "postDate": "2019-01-30T02:28:05.490Z", "content": "very interesting thinking and so dirty, because it will not affect the normal progress of the competition, I will not set it to cheating.
", "rawMarkdown": "very interesting thinking and so dirty, because it will not affect the normal progress of the competition, I will not set it to cheating.", "votes": 2 }, { "id": 465832, "postDate": "2019-02-04T05:13:12.607Z", "content": "How much, in general, this information can help? \nKernels will be rerun anyways with a different dataset.
", "rawMarkdown": "How much, in general, this information can help? \nKernels will be rerun anyways with a different dataset." }, { "id": 465324, "postDate": "2019-02-02T20:26:08.860Z", "content": "Nice
", "rawMarkdown": "Nice" }, { "id": 465225, "postDate": "2019-02-02T16:48:22.300Z", "content": "respect well done!!, we need more people like you here. These things are out of control since long ago.
", "rawMarkdown": "*respect* well done!!, we need more people like you here. These things are out of control since long ago." }, { "id": 465059, "postDate": "2019-02-02T08:32:03.070Z", "content": "got it
", "rawMarkdown": "got it" }, { "id": 465026, "postDate": "2019-02-02T06:02:24.893Z", "content": "Nice :D
", "rawMarkdown": "Nice :D" }, { "id": 464396, "postDate": "2019-01-31T19:41:58.513Z", "content": "Nice Work!
", "rawMarkdown": "Nice Work!" }, { "id": 464260, "postDate": "2019-01-31T13:49:17.173Z", "content": "Nice trick.
", "rawMarkdown": "Nice trick." }, { "id": 464243, "postDate": "2019-01-31T13:04:21.027Z", "content": "First, we need the video :P ... Next now instead of having 50 accuracies as the baseline classifier, we can have this as the baseline XD
", "rawMarkdown": "First, we need the video :P ... Next now instead of having 50 accuracies as the baseline classifier, we can have this as the baseline XD" }, { "id": 463871, "postDate": "2019-01-30T19:32:06.957Z", "content": "interesting
", "rawMarkdown": "interesting" }, { "id": 463800, "postDate": "2019-01-30T16:28:46.773Z", "content": "Nice
", "rawMarkdown": "Nice" }, { "id": 463599, "postDate": "2019-01-30T09:08:49.193Z", "content": "it's funny, enjoy it.
", "rawMarkdown": "it's funny, enjoy it. " }, { "id": 463308, "postDate": "2019-01-29T19:32:14.573Z", "content": "Nice
", "rawMarkdown": "Nice" }, { "id": 465450, "postDate": "2019-02-03T05:59:02.787Z", "rawMarkdown": "", "isDeleted": true }, { "id": 464050, "postDate": "2019-01-31T05:55:32.537Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 464057, "postDate": "2019-01-31T06:07:16.833Z", "content": "maybe the data hanging on https://pastebin.com is also be published by kaggler.
", "rawMarkdown": "maybe the data hanging on https://pastebin.com is also be published by kaggler." } ] }, { "id": 463448, "postDate": "2019-01-30T02:05:17.380Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 463562, "postDate": "2019-01-30T07:29:25.280Z", "content": "what exactly? leaving the data hanging on https://pastebin.com or the score?
", "rawMarkdown": "what exactly? leaving the data hanging on https://pastebin.com or the score?", "votes": -1 }, { "id": 464658, "postDate": "2019-02-01T08:39:57.017Z", "content": "good work pavel
", "rawMarkdown": "good work pavel", "votes": 1 } ] }, { "id": 464534, "postDate": "2019-02-01T03:31:21.823Z", "content": "Thank pro!
", "rawMarkdown": "Thank pro!" }, { "id": 463697, "postDate": "2019-01-30T12:54:40.493Z", "content": "Thanks a lot
", "rawMarkdown": "Thanks a lot\n" } ], "comments": [ { "id": 463158, "author_name": "Shahebaz Mohammad", "author_url": "", "post_date": "2019-01-29T14:48:42.567000", "content": "I mark this trick as Insincere :)
", "votes": 10, "replies": [ { "id": 463171, "author_name": "Shahebaz Mohammad", "author_url": "", "post_date": "2019-01-29T15:12:54.487000", "content": "On the second thoughts, I was wondering how a submission ended up there. This potentially means offline teaming or solution sharing of some kind. Puts a real question in my head;
\n\nthere are no reasons to trust public LB at this competition, regardless of this submission
", "votes": 8, "replies": [] }, { "id": 463203, "author_name": "Shahebaz Mohammad", "author_url": "", "post_date": "2019-01-29T15:56:44.177000", "content": "True. I agree
", "votes": -3, "replies": [] }, { "id": 463963, "author_name": "Kiniggit", "author_url": "", "post_date": "2019-01-31T00:48:58.337000", "content": "for newcomers like me, LB just gets you anxious! Paulo keeps reminding me to forget LB... Easier said than done...
", "votes": 1, "replies": [] } ] }, { "id": 463361, "author_name": "Yury Kashnitsky", "author_url": "", "post_date": "2019-01-29T21:29:42.013000", "content": "I don't know what you all are upvoting here, folks. Would you like such dirty hacks to be well-known and have many more dummy kagglers on the leaderboard?
", "votes": 10, "replies": [ { "id": 463561, "author_name": "", "author_url": "", "post_date": "2019-01-30T07:28:06.357000", "content": "the post actually was about not leaving your private data on https://pastebin.com
", "votes": 5, "replies": [] }, { "id": 463581, "author_name": "Yury Kashnitsky", "author_url": "", "post_date": "2019-01-30T08:22:02.853000", "content": "Of course I got it. But at the same time for many others the post can serve as a call to cheating.
", "votes": 2, "replies": [] }, { "id": 464659, "author_name": "Muhammad Tahir A.", "author_url": "", "post_date": "2019-02-01T08:41:33.633000", "content": "Yury we should appreciate everyone
", "votes": 1, "replies": [] } ] }, { "id": 462990, "author_name": "Aditya Soni", "author_url": "", "post_date": "2019-01-29T07:46:17.523000", "content": "Really a very Dirty Trick :)
", "votes": 3, "replies": [] }, { "id": 463163, "author_name": "Lyalikov Artyom", "author_url": "", "post_date": "2019-01-29T14:54:46.043000", "content": "True detective :D
", "votes": 4, "replies": [ { "id": 463196, "author_name": "", "author_url": "", "post_date": "2019-01-29T15:41:23.990000", "content": "that what makes Kaggle so fun for me, what gives me drive
", "votes": 4, "replies": [] } ] }, { "id": 463098, "author_name": "大黑楼", "author_url": "", "post_date": "2019-01-29T11:51:51.287000", "content": "no use
", "votes": 3, "replies": [] }, { "id": 464657, "author_name": "Muhammad Tahir A.", "author_url": "", "post_date": "2019-02-01T08:38:51.137000", "content": "Thnx We know
", "votes": 1, "replies": [] }, { "id": 463474, "author_name": "Shujian Liu", "author_url": "", "post_date": "2019-01-30T03:33:02.783000", "content": "Probably this is useful to check where our models predict wrong and fix the model a little bit ....
", "votes": 1, "replies": [] }, { "id": 463454, "author_name": "Bai", "author_url": "", "post_date": "2019-01-30T02:28:05.490000", "content": "very interesting thinking and so dirty, because it will not affect the normal progress of the competition, I will not set it to cheating.
", "votes": 2, "replies": [] }, { "id": 465832, "author_name": "lizzzi1", "author_url": "", "post_date": "2019-02-04T05:13:12.607000", "content": "How much, in general, this information can help? \nKernels will be rerun anyways with a different dataset.
", "votes": 0, "replies": [] }, { "id": 465324, "author_name": "Mark S. Shenouda", "author_url": "", "post_date": "2019-02-02T20:26:08.860000", "content": "Nice
", "votes": 0, "replies": [] }, { "id": 465225, "author_name": "NxGTR", "author_url": "", "post_date": "2019-02-02T16:48:22.300000", "content": "respect well done!!, we need more people like you here. These things are out of control since long ago.
", "votes": 0, "replies": [] }, { "id": 465059, "author_name": "vishal shrivastava", "author_url": "", "post_date": "2019-02-02T08:32:03.070000", "content": "got it
", "votes": 0, "replies": [] }, { "id": 465026, "author_name": "shreyassv", "author_url": "", "post_date": "2019-02-02T06:02:24.893000", "content": "Nice :D
", "votes": 0, "replies": [] }, { "id": 464396, "author_name": "Akshat Singhal", "author_url": "", "post_date": "2019-01-31T19:41:58.513000", "content": "Nice Work!
", "votes": 0, "replies": [] }, { "id": 464260, "author_name": "Adithya", "author_url": "", "post_date": "2019-01-31T13:49:17.173000", "content": "Nice trick.
", "votes": 0, "replies": [] }, { "id": 464243, "author_name": "Samim Ekram", "author_url": "", "post_date": "2019-01-31T13:04:21.027000", "content": "First, we need the video :P ... Next now instead of having 50 accuracies as the baseline classifier, we can have this as the baseline XD
", "votes": 0, "replies": [] }, { "id": 463871, "author_name": "Karsten", "author_url": "", "post_date": "2019-01-30T19:32:06.957000", "content": "interesting
", "votes": 0, "replies": [] }, { "id": 463800, "author_name": "Alex Jenkins", "author_url": "", "post_date": "2019-01-30T16:28:46.773000", "content": "Nice
", "votes": 0, "replies": [] }, { "id": 463599, "author_name": "Chin", "author_url": "", "post_date": "2019-01-30T09:08:49.193000", "content": "it's funny, enjoy it.
", "votes": 0, "replies": [] }, { "id": 463308, "author_name": "RahulSingh", "author_url": "", "post_date": "2019-01-29T19:32:14.573000", "content": "Nice
", "votes": 0, "replies": [] }, { "id": 465450, "author_name": "", "author_url": "", "post_date": "2019-02-03T05:59:02.787000", "content": "", "votes": 0, "replies": [] }, { "id": 464050, "author_name": "", "author_url": "", "post_date": "2019-01-31T05:55:32.537000", "content": "", "votes": 0, "replies": [ { "id": 464057, "author_name": "Bai", "author_url": "", "post_date": "2019-01-31T06:07:16.833000", "content": "maybe the data hanging on https://pastebin.com is also be published by kaggler.
", "votes": 0, "replies": [] } ] }, { "id": 463448, "author_name": "", "author_url": "", "post_date": "2019-01-30T02:05:17.380000", "content": "", "votes": 0, "replies": [ { "id": 463562, "author_name": "", "author_url": "", "post_date": "2019-01-30T07:29:25.280000", "content": "what exactly? leaving the data hanging on https://pastebin.com or the score?
", "votes": -1, "replies": [] }, { "id": 464658, "author_name": "Muhammad Tahir A.", "author_url": "", "post_date": "2019-02-01T08:39:57.017000", "content": "good work pavel
", "votes": 1, "replies": [] } ] }, { "id": 464534, "author_name": "An An", "author_url": "", "post_date": "2019-02-01T03:31:21.823000", "content": "Thank pro!
", "votes": 0, "replies": [] }, { "id": 463697, "author_name": "Iurii Cojocari", "author_url": "", "post_date": "2019-01-30T12:54:40.493000", "content": "Thanks a lot
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "462981": "As you might know, a wonderful event was held this weekend - Kaggledays Paris. I was doing a workshop there called \"Dirty tricks for image classification\" (video should be available soon).\n\nAmong several tricks I shared with the participants was googling file names from the test set. Guess what? It also works for text classification :)\n\nTry to google *00c660c58101eafe25e9*. There is going to be only one result: this link https://pastebin.com/KeVgnB6L which looks very much like a submission. To save your attempts, it scores 0.699 on LB. It was uploaded on NOV 16TH, 2018 - the exact same date when someone called **lcch** got the same result (see [raw data][1]).\n\nAnother day, another \"leak\"... another valuable lesson for @lcch135616 ;)\n\n [1]: https://www.kaggle.com/c/10737/publicleaderboarddata.zip", "463158": "I mark this trick as **Insincere** :) ", "463361": "I don't know what you all are upvoting here, folks. Would you like such dirty hacks to be well-known and have many more dummy kagglers on the leaderboard? ", "462990": "Really a very Dirty Trick :)", "463163": "True detective :D", "463098": "no use", "464657": "Thnx We know", "463474": "Probably this is useful to check where our models predict wrong and fix the model a little bit ....", "463454": "very interesting thinking and so dirty, because it will not affect the normal progress of the competition, I will not set it to cheating.", "465832": "How much, in general, this information can help? \nKernels will be rerun anyways with a different dataset.", "465324": "Nice", "465225": "*respect* well done!!, we need more people like you here. These things are out of control since long ago.", "465059": "got it", "465026": "Nice :D", "464396": "Nice Work!", "464260": "Nice trick.", "464243": "First, we need the video :P ... Next now instead of having 50 accuracies as the baseline classifier, we can have this as the baseline XD", "463871": "interesting", "463800": "Nice", "463599": "it's funny, enjoy it. ", "463308": "Nice", "465450": "", "464050": "", "463448": "", "464534": "Thank pro!", "463697": "Thanks a lot\n" } }