Probably the most redundant competition results when sample submission of 0.5,0.5 gets a bronze medal eh?
\nDid anyone get 0.5 in their unselected scores?
", "messages": [ { "id": 1973901, "postDate": "2022-10-06T00:02:12.270Z", "content": "Probably the most redundant competition results when sample submission of 0.5,0.5 gets a bronze medal eh?
\nDid anyone get 0.5 in their unselected scores?
", "rawMarkdown": "Probably the most redundant competition results when sample submission of 0.5,0.5 gets a bronze medal eh?\n\nDid anyone get 0.5 in their unselected scores?", "votes": 31 }, { "id": 1976609, "postDate": "2022-10-07T13:02:03.503Z", "content": "Hearty congratulations to the sample submission 👍for the medal!
", "rawMarkdown": "Hearty congratulations to the sample submission 👍for the medal!", "votes": 5 }, { "id": 1974528, "postDate": "2022-10-06T09:45:14.940Z", "content": "\"That means not many of us provided a good method for this task.\"
\ni think it is problem with the data … perhaps there isn't enough information in the data itself.
", "rawMarkdown": "\"That means not many of us provided a good method for this task.\"\n\ni think it is problem with the data ... perhaps there isn't enough information in the data itself.", "votes": 5, "replies": [ { "id": 1974543, "postDate": "2022-10-06T09:56:06.093Z", "content": "I guess most of the people understood the problem statement wrong. In my opinion, the problem was to create a model that is able to distinguish between classes without being bias towards them even after class imbalance in the training dataset. By fixing this issue with my end submission, i was able to get 0.6 private lb score from resnet50 model which i think could have been improved if more complex models were applied and really had the time for the same.
", "rawMarkdown": "I guess most of the people understood the problem statement wrong. In my opinion, the problem was to create a model that is able to distinguish between classes without being bias towards them even after class imbalance in the training dataset. By fixing this issue with my end submission, i was able to get 0.6 private lb score from resnet50 model which i think could have been improved if more complex models were applied and really had the time for the same. ", "votes": 2 }, { "id": 1974544, "postDate": "2022-10-06T09:56:36.747Z", "content": "Also i feel like tiling in this case really helped. Would love to know yours and others feedback
", "rawMarkdown": "Also i feel like tiling in this case really helped. Would love to know yours and others feedback" }, { "id": 1985986, "postDate": "2022-10-13T18:21:52.550Z", "content": "100% it is the data. Just poor data overall, would have to work a miracle for actually useful results
", "rawMarkdown": "100% it is the data. Just poor data overall, would have to work a miracle for actually useful results" } ] }, { "id": 1973917, "postDate": "2022-10-06T00:15:47.177Z", "content": "Wow, sample submission gets bronze medal lol
", "rawMarkdown": "Wow, sample submission gets bronze medal lol", "votes": 6 }, { "id": 1973948, "postDate": "2022-10-06T01:12:25.067Z", "content": "I wonder if the competition's hosts threw away all their money on this competition. Absolutely random results.
", "rawMarkdown": "I wonder if the competition's hosts threw away all their money on this competition. Absolutely random results. ", "votes": 3 }, { "id": 1976535, "postDate": "2022-10-07T12:11:00.997Z", "content": "This is bonkers… that said:
\nI think this is indicative of how easy it was to overfit the training data when using the public LB to gauge success.
", "rawMarkdown": "This is bonkers… that said:\n\nI think this is indicative of how easy it was to overfit the training data when using the public LB to gauge success.", "votes": 4 }, { "id": 1975935, "postDate": "2022-10-07T05:14:03.120Z", "content": "What an amazing achievement. I didn't able to submit my notebook.
", "rawMarkdown": "What an amazing achievement. I didn't able to submit my notebook.", "votes": 1 }, { "id": 1975568, "postDate": "2022-10-06T21:19:57.093Z", "content": "I wish i made one as I got so bored of constant fails 🤕🤒
", "rawMarkdown": "I wish i made one as I got so bored of constant fails 🤕🤒", "votes": 1 }, { "id": 1975550, "postDate": "2022-10-06T20:40:50.163Z", "content": "I tried imagenet at last moment but could not submit as it took so much time and ran out of memory so my basic model must have been selected. Happy for my score though. Could have improved if started earlier.
", "rawMarkdown": "I tried imagenet at last moment but could not submit as it took so much time and ran out of memory so my basic model must have been selected. Happy for my score though. Could have improved if started earlier. ", "votes": 1 }, { "id": 1974057, "postDate": "2022-10-06T04:01:01.637Z", "content": "I have a good plan. Unfortunately, it has timed out
", "rawMarkdown": "I have a good plan. Unfortunately, it has timed out", "votes": 1 }, { "id": 1973911, "postDate": "2022-10-06T00:12:28.887Z", "content": "public 0.5 -> private 1.3
\n:'(
Well I'm very happy. I made 23 fail submissions but the last one yesterday was successful 😂😂😂. It is a pitty that the scoring takes too much hours and ended up an hour after the deadline.
", "rawMarkdown": "Well I'm very happy. I made 23 fail submissions but the last one yesterday was successful 😂😂😂. It is a pitty that the scoring takes too much hours and ended up an hour after the deadline.", "votes": 2 }, { "id": 1974053, "postDate": "2022-10-06T03:58:19.370Z", "content": "It doesn't necessarily mean useless. I take it as hardest.
", "rawMarkdown": "It doesn't necessarily mean useless. I take it as hardest.", "votes": 2 }, { "id": 1973909, "postDate": "2022-10-06T00:07:17.637Z", "content": "I enjoyed my 7 days journey in this competition, I had the chance to feel the fake gold for a while hhhh
\nThis was the best fall ever … and yes I envy sample_submission 😒
Damn that was a hard fall 0.4 -> 0.9 (796)😯. I hope you get your gold in another comp.
", "rawMarkdown": "Damn that was a hard fall 0.4 -> 0.9 (796)😯. I hope you get your gold in another comp." }, { "id": 1974426, "postDate": "2022-10-06T08:41:26.187Z", "content": "Its ok , I didn’t put efforts on this one, and I was sure that I will fall in PB.
\nIt was a lucky shot :-)
", "rawMarkdown": "Its ok , I didn’t put efforts on this one, and I was sure that I will fall in PB.\n\nIt was a lucky shot :-)" }, { "id": 1975041, "postDate": "2022-10-06T15:31:44.700Z", "content": "@asalhi you did help by confirming that your high scoring submission was not your best model. You deserve credit for that, thanks again and all the best for future competitions :)
", "rawMarkdown": "@asalhi you did help by confirming that your high scoring submission was not your best model. You deserve credit for that, thanks again and all the best for future competitions :)", "votes": 1 } ] }, { "id": 1976483, "postDate": "2022-10-07T11:41:29.910Z", "content": "Congratulations to
\nNull Pointer Exception:
\nplayer name is undefined!
Just joking, this was quite an interesting outcome, I would have never expected this!
", "rawMarkdown": "Congratulations to \n\nNull Pointer Exception:\nplayer name is undefined! \n\nJust joking, this was quite an interesting outcome, I would have never expected this!\n" }, { "id": 1973922, "postDate": "2022-10-06T00:25:54.580Z", "content": "That means not many of us provided a good method for this task. And about 3% of the test set was for public score, so massive shake was too predictable. Please don't make everyone's effort useless. For me, it was a good opportunity to learn how to deal with large scale datasets, and concept of multi instance learning.
", "rawMarkdown": "That means not many of us provided a good method for this task. And about 3% of the test set was for public score, so massive shake was too predictable. Please don't make everyone's effort useless. For me, it was a good opportunity to learn how to deal with large scale datasets, and concept of multi instance learning. ", "votes": 8, "isDeleted": true, "replies": [ { "id": 1973938, "postDate": "2022-10-06T00:49:37.330Z", "content": "Agreed! Most of my struggle was trying to handle the large images without running out of memory or in a way that isn't slow to process. I am interested to see if anyone came up with any innovative approaches. It was a fun competition to apart of and learn.
", "rawMarkdown": "Agreed! Most of my struggle was trying to handle the large images without running out of memory or in a way that isn't slow to process. I am interested to see if anyone came up with any innovative approaches. It was a fun competition to apart of and learn.", "votes": 2 }, { "id": 1974400, "postDate": "2022-10-06T08:30:48.813Z", "content": "\n\nPlease don't make everyone's effort useless
\n
Was specifically referring to results, I am curious if the hosts are pleased with the outcome, maybe it is good learning that we could not greatly surpass equal weighted predictions.
\nI hope everyone still learned a lot of course..
", "rawMarkdown": ">Please don't make everyone's effort useless\n\nWas specifically referring to results, I am curious if the hosts are pleased with the outcome, maybe it is good learning that we could not greatly surpass equal weighted predictions.\n\nI hope everyone still learned a lot of course.." } ] } ], "comments": [ { "id": 1976609, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2022-10-07T13:02:03.503000", "content": "Hearty congratulations to the sample submission 👍for the medal!
", "votes": 5, "replies": [] }, { "id": 1974528, "author_name": "hengck23", "author_url": "", "post_date": "2022-10-06T09:45:14.940000", "content": "\"That means not many of us provided a good method for this task.\"
\ni think it is problem with the data … perhaps there isn't enough information in the data itself.
", "votes": 5, "replies": [ { "id": 1974543, "author_name": "Mrinal Tyagi", "author_url": "", "post_date": "2022-10-06T09:56:06.093000", "content": "I guess most of the people understood the problem statement wrong. In my opinion, the problem was to create a model that is able to distinguish between classes without being bias towards them even after class imbalance in the training dataset. By fixing this issue with my end submission, i was able to get 0.6 private lb score from resnet50 model which i think could have been improved if more complex models were applied and really had the time for the same.
", "votes": 2, "replies": [] }, { "id": 1974544, "author_name": "Mrinal Tyagi", "author_url": "", "post_date": "2022-10-06T09:56:36.747000", "content": "Also i feel like tiling in this case really helped. Would love to know yours and others feedback
", "votes": 0, "replies": [] }, { "id": 1985986, "author_name": "Iuryck Santos", "author_url": "", "post_date": "2022-10-13T18:21:52.550000", "content": "100% it is the data. Just poor data overall, would have to work a miracle for actually useful results
", "votes": 0, "replies": [] } ] }, { "id": 1973917, "author_name": "Chris Deotte", "author_url": "", "post_date": "2022-10-06T00:15:47.177000", "content": "Wow, sample submission gets bronze medal lol
", "votes": 6, "replies": [] }, { "id": 1973948, "author_name": "moth", "author_url": "", "post_date": "2022-10-06T01:12:25.067000", "content": "I wonder if the competition's hosts threw away all their money on this competition. Absolutely random results.
", "votes": 3, "replies": [] }, { "id": 1976535, "author_name": "Darien Schettler", "author_url": "", "post_date": "2022-10-07T12:11:00.997000", "content": "This is bonkers… that said:
\nI think this is indicative of how easy it was to overfit the training data when using the public LB to gauge success.
", "votes": 4, "replies": [] }, { "id": 1975935, "author_name": "Rahul Pandey", "author_url": "", "post_date": "2022-10-07T05:14:03.120000", "content": "What an amazing achievement. I didn't able to submit my notebook.
", "votes": 1, "replies": [] }, { "id": 1975568, "author_name": "Muhammad Ammar Jamshed", "author_url": "", "post_date": "2022-10-06T21:19:57.093000", "content": "I wish i made one as I got so bored of constant fails 🤕🤒
", "votes": 1, "replies": [] }, { "id": 1975550, "author_name": "Chirag Desai", "author_url": "", "post_date": "2022-10-06T20:40:50.163000", "content": "I tried imagenet at last moment but could not submit as it took so much time and ran out of memory so my basic model must have been selected. Happy for my score though. Could have improved if started earlier.
", "votes": 1, "replies": [] }, { "id": 1974057, "author_name": "qiucen", "author_url": "", "post_date": "2022-10-06T04:01:01.637000", "content": "I have a good plan. Unfortunately, it has timed out
", "votes": 1, "replies": [] }, { "id": 1973911, "author_name": "Kamo Neggi", "author_url": "", "post_date": "2022-10-06T00:12:28.887000", "content": "public 0.5 -> private 1.3
\n:'(
Well I'm very happy. I made 23 fail submissions but the last one yesterday was successful 😂😂😂. It is a pitty that the scoring takes too much hours and ended up an hour after the deadline.
", "votes": 2, "replies": [] }, { "id": 1974053, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2022-10-06T03:58:19.370000", "content": "It doesn't necessarily mean useless. I take it as hardest.
", "votes": 2, "replies": [] }, { "id": 1973909, "author_name": "Ali", "author_url": "", "post_date": "2022-10-06T00:07:17.637000", "content": "I enjoyed my 7 days journey in this competition, I had the chance to feel the fake gold for a while hhhh
\nThis was the best fall ever … and yes I envy sample_submission 😒
Damn that was a hard fall 0.4 -> 0.9 (796)😯. I hope you get your gold in another comp.
", "votes": 0, "replies": [] }, { "id": 1974426, "author_name": "Ali", "author_url": "", "post_date": "2022-10-06T08:41:26.187000", "content": "Its ok , I didn’t put efforts on this one, and I was sure that I will fall in PB.
\nIt was a lucky shot :-)
", "votes": 0, "replies": [] }, { "id": 1975041, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-06T15:31:44.700000", "content": "@asalhi you did help by confirming that your high scoring submission was not your best model. You deserve credit for that, thanks again and all the best for future competitions :)
", "votes": 1, "replies": [] } ] }, { "id": 1976483, "author_name": "Pietro Maldini", "author_url": "", "post_date": "2022-10-07T11:41:29.910000", "content": "Congratulations to
\nNull Pointer Exception:
\nplayer name is undefined!
Just joking, this was quite an interesting outcome, I would have never expected this!
", "votes": 0, "replies": [] }, { "id": 1973922, "author_name": "", "author_url": "", "post_date": "2022-10-06T00:25:54.580000", "content": "That means not many of us provided a good method for this task. And about 3% of the test set was for public score, so massive shake was too predictable. Please don't make everyone's effort useless. For me, it was a good opportunity to learn how to deal with large scale datasets, and concept of multi instance learning.
", "votes": 8, "replies": [ { "id": 1973938, "author_name": "", "author_url": "", "post_date": "2022-10-06T00:49:37.330000", "content": "Agreed! Most of my struggle was trying to handle the large images without running out of memory or in a way that isn't slow to process. I am interested to see if anyone came up with any innovative approaches. It was a fun competition to apart of and learn.
", "votes": 2, "replies": [] }, { "id": 1974400, "author_name": "JM", "author_url": "", "post_date": "2022-10-06T08:30:48.813000", "content": "\n\nPlease don't make everyone's effort useless
\n
Was specifically referring to results, I am curious if the hosts are pleased with the outcome, maybe it is good learning that we could not greatly surpass equal weighted predictions.
\nI hope everyone still learned a lot of course..
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1973901": "Probably the most redundant competition results when sample submission of 0.5,0.5 gets a bronze medal eh?\n\nDid anyone get 0.5 in their unselected scores?", "1976609": "Hearty congratulations to the sample submission 👍for the medal!", "1974528": "\"That means not many of us provided a good method for this task.\"\n\ni think it is problem with the data ... perhaps there isn't enough information in the data itself.", "1973917": "Wow, sample submission gets bronze medal lol", "1973948": "I wonder if the competition's hosts threw away all their money on this competition. Absolutely random results. ", "1976535": "This is bonkers… that said:\n\nI think this is indicative of how easy it was to overfit the training data when using the public LB to gauge success.", "1975935": "What an amazing achievement. I didn't able to submit my notebook.", "1975568": "I wish i made one as I got so bored of constant fails 🤕🤒", "1975550": "I tried imagenet at last moment but could not submit as it took so much time and ran out of memory so my basic model must have been selected. Happy for my score though. Could have improved if started earlier. ", "1974057": "I have a good plan. Unfortunately, it has timed out", "1973911": "public 0.5 -> private 1.3\n:'(", "1974584": "Well I'm very happy. I made 23 fail submissions but the last one yesterday was successful 😂😂😂. It is a pitty that the scoring takes too much hours and ended up an hour after the deadline.", "1974053": "It doesn't necessarily mean useless. I take it as hardest.", "1973909": "I enjoyed my 7 days journey in this competition, I had the chance to feel the fake gold for a while hhhh\nThis was the best fall ever … and yes I envy sample_submission 😒", "1976483": "Congratulations to \n\nNull Pointer Exception:\nplayer name is undefined! \n\nJust joking, this was quite an interesting outcome, I would have never expected this!\n", "1973922": "That means not many of us provided a good method for this task. And about 3% of the test set was for public score, so massive shake was too predictable. Please don't make everyone's effort useless. For me, it was a good opportunity to learn how to deal with large scale datasets, and concept of multi instance learning. " } }