I got some helpful feedback so I'm editing this post. Thanks for that. Basically I didn't have time for this competition, saw it was very unstable, and submitted two public kernel ensembles.
\nI found sometimes in these very unstable competitions, there are alot of public overfitting that may not perform well in private LB, and sometimes a heavily ensembled solution ends up being stable enough to weather the shakeup. The submissions were one ensemble of the top 9 public lb kernels and the other one about the top ~20 public kernels. The top ~20 lb kernels blend scored 32nd place. It just used straight average.
\nI did this before for the ASHRAE competition and I silvered that one as well. That ASHRAE competition was also unstable with crazy public overfitting if I remember correctly.
\nI originally posted encase someone found it interesting.
\nThis post has been edited.
", "messages": [ { "id": 975029, "postDate": "2020-08-18T05:52:17.967Z", "content": "First of all congrats on your high silver medal . Good for you . I have not downvoted you as I didn't want to bring sadness in your joy .
\nHowever there are few things that I do find a bit off :
\na) I have got a medal once just by tweaking a public model and submitting few hours before competition end . Was that illegal ? No . But can I promote that behaviour ? No . Because , I will then be misguiding a lot of people who probably would look upto kaggle community to learn . It was mere luck , there was no fun , I didn't learn anything .
\nb) Is ensembling bad ? No . But along with your post if you would have put the factors if any to chose model for ensemble , method of ensemble then probably people would also know that what are the tricks of the trade rather than probably blindly following a practice .
I think probably few people found these small points worth to downvote .
", "rawMarkdown": "First of all congrats on your high silver medal . Good for you . I have not downvoted you as I didn't want to bring sadness in your joy . \n\nHowever there are few things that I do find a bit off :\na) I have got a medal once just by tweaking a public model and submitting few hours before competition end . Was that illegal ? No . But can I promote that behaviour ? No . Because , I will then be misguiding a lot of people who probably would look upto kaggle community to learn . It was mere luck , there was no fun , I didn't learn anything . \nb) Is ensembling bad ? No . But along with your post if you would have put the factors if any to chose model for ensemble , method of ensemble then probably people would also know that what are the tricks of the trade rather than probably blindly following a practice . \n\nI think probably few people found these small points worth to downvote . ", "votes": 10 }, { "id": 974586, "postDate": "2020-08-18T01:46:07.647Z", "content": "I got some helpful feedback so I'm editing this post. Thanks for that. Basically I didn't have time for this competition, saw it was very unstable, and submitted two public kernel ensembles.
\nI found sometimes in these very unstable competitions, there are alot of public overfitting that may not perform well in private LB, and sometimes a heavily ensembled solution ends up being stable enough to weather the shakeup. The submissions were one ensemble of the top 9 public lb kernels and the other one about the top ~20 public kernels. The top ~20 lb kernels blend scored 32nd place. It just used straight average.
\nI did this before for the ASHRAE competition and I silvered that one as well. That ASHRAE competition was also unstable with crazy public overfitting if I remember correctly.
\nI originally posted encase someone found it interesting.
\nThis post has been edited.
", "rawMarkdown": "I got some helpful feedback so I'm editing this post. Thanks for that. Basically I didn't have time for this competition, saw it was very unstable, and submitted two public kernel ensembles. \n\nI found sometimes in these very unstable competitions, there are alot of public overfitting that may not perform well in private LB, and sometimes a heavily ensembled solution ends up being stable enough to weather the shakeup. The submissions were one ensemble of the top 9 public lb kernels and the other one about the top ~20 public kernels. The top ~20 lb kernels blend scored 32nd place. It just used straight average. \n\nI did this before for the ASHRAE competition and I silvered that one as well. That ASHRAE competition was also unstable with crazy public overfitting if I remember correctly.\n\nI originally posted encase someone found it interesting. \n\nThis post has been edited.", "votes": 16 }, { "id": 975803, "postDate": "2020-08-18T13:26:14.140Z", "content": "First good for you, second please be careful about how you phrase this: I did nothing. Because some us did work hard to come out with our solution, which may not be as good as blindly blending 20 public kernels but which has the taste of sweat, authenticity and hard work. To be honest I don't even understand why someone would spend time blending 20 public kernels but that's your problem not mine.
On a more interesting note, what happens if one of the public notebooks does not follow the competition rules? People should not be able to use it, but how would Kaggle know? This is an interesting question in my opinion and your example shows that public sharing can become an issue even in top 1-2% of Kaggle Competitions.
\nAs an example : Imagine someone hand labeled their test solution and called it a blend of blend. Then what happens to people who use it? Everyone is disqualified? How do you spot them?
", "rawMarkdown": "First good for you, second please be careful about how you phrase this: `I did nothing`. Because some us did work hard to come out with our solution, which may not be as good as blindly blending 20 public kernels but which has the taste of sweat, authenticity and hard work. To be honest I don't even understand why someone would spend time blending 20 public kernels but that's your problem not mine.\n\nOn a more interesting note, what happens if one of the public notebooks does not follow the competition rules? People should not be able to use it, but how would Kaggle know? This is an interesting question in my opinion and your example shows that public sharing can become an issue even in top 1-2% of Kaggle Competitions.\n\nAs an example : Imagine someone hand labeled their test solution and called it a blend of blend. Then what happens to people who use it? Everyone is disqualified? How do you spot them?\n\n\n", "votes": 7 }, { "id": 975077, "postDate": "2020-08-18T06:15:50.060Z", "content": "Your arrogant bragging with getting a silver medal by doing nothing is a slap in the face for everybody that worked hard.
", "rawMarkdown": "Your arrogant bragging with getting a silver medal by doing nothing is a slap in the face for everybody that worked hard. ", "votes": 8, "replies": [ { "id": 975098, "postDate": "2020-08-18T06:37:52.787Z", "rawMarkdown": "", "votes": -3, "isDeleted": true } ] }, { "id": 974651, "postDate": "2020-08-18T02:17:31.940Z", "content": "Congrats! So what kind of blending method did you use? I find postprocessing is quite important as well.
", "rawMarkdown": "Congrats! So what kind of blending method did you use? I find postprocessing is quite important as well.", "votes": 3 }, { "id": 974622, "postDate": "2020-08-18T02:03:08.043Z", "content": "Thanks for your honesty !
\nI did all the work by myself (refuse to use public kernels) --> Look where I got 8-(
", "rawMarkdown": "Thanks for your honesty !\n\nI did all the work by myself (refuse to use public kernels) --> Look where I got 8-(", "votes": 3, "replies": [ { "id": 974637, "postDate": "2020-08-18T02:09:15.377Z", "content": "It's fine its much fun to work on our own cv then to submit solutions which you don't even understand
", "rawMarkdown": "It's fine its much fun to work on our own cv then to submit solutions which you don't even understand ", "votes": 1 } ] }, { "id": 974629, "postDate": "2020-08-18T02:05:47.163Z", "content": "Thanks for your honesty - appreciate it. :)
", "rawMarkdown": "Thanks for your honesty - appreciate it. :) ", "votes": 4 }, { "id": 974626, "postDate": "2020-08-18T02:04:52.680Z", "content": "I'm just sharing what I did. I didn't have much time for Kaggle, saw this competition is very unstable, and submitted top ~20 blend. And got 34rd. I'm just sharing it encase someone finds interesting.
", "rawMarkdown": "I'm just sharing what I did. I didn't have much time for Kaggle, saw this competition is very unstable, and submitted top ~20 blend. And got 34rd. I'm just sharing it encase someone finds interesting. ", "votes": 3, "replies": [ { "id": 974631, "postDate": "2020-08-18T02:06:20.010Z", "content": "It means that you have great skills in model blending! :D. Thank you for your honesty!
", "rawMarkdown": "It means that you have great skills in model blending! :D. Thank you for your honesty!", "votes": 1 } ] }, { "id": 974627, "postDate": "2020-08-18T02:05:21.093Z", "content": "Why the downvotes? Anyone cares to elaborate?
", "rawMarkdown": "Why the downvotes? Anyone cares to elaborate?", "votes": 2 }, { "id": 975719, "postDate": "2020-08-18T12:44:18.033Z", "content": "So you’re saying if I were to do this now - take 20 top public kernels and do simple average, I’d get rank 43 score? :)
", "rawMarkdown": "So you’re saying if I were to do this now - take 20 top public kernels and do simple average, I’d get rank 43 score? :)", "votes": 1 }, { "id": 975085, "postDate": "2020-08-18T06:20:19.073Z", "content": "M5 Deja vu. It's the fault of hosts, not yours, you shouldn't be ashamed of that.
", "rawMarkdown": "M5 Deja vu. It's the fault of hosts, not yours, you shouldn't be ashamed of that.", "votes": 1 }, { "id": 974898, "postDate": "2020-08-18T04:46:01.750Z", "content": "Congrats for winning a silver medal.
", "rawMarkdown": "Congrats for winning a silver medal.", "votes": 1 }, { "id": 974800, "postDate": "2020-08-18T03:58:06.317Z", "content": "Some of the high scoring public notebooks had good private scores like public 0.9605 and private 0.9446 which could have got you in top 40. On the other hand, some were really bad like public 0.9619 to private 0.9128. So yeah, luck is indeed a factor in the private leaderboard.😄
\nCongrats on your medal 💯
Appreciate your honesty @bopengiowa .. Congrats on your silver..
", "rawMarkdown": "Appreciate your honesty @bopengiowa .. Congrats on your silver.. ", "votes": 1 }, { "id": 974655, "postDate": "2020-08-18T02:19:19.743Z", "content": "LOL
\nWould you mind sharing how you blended those models? Any special blending techniques??
", "rawMarkdown": "LOL \n\nWould you mind sharing how you blended those models? Any special blending techniques??", "votes": 1 }, { "id": 974590, "postDate": "2020-08-18T01:48:58.457Z", "content": "Interesting! LOL
", "rawMarkdown": "Interesting! LOL", "votes": 1 }, { "id": 979184, "postDate": "2020-08-20T17:19:40.733Z", "content": "Thanks for the feedback. It helps me learn what people feel about various things. I saw the feedback and tried to modify the post to cater to your desires and recommendations.
", "rawMarkdown": "Thanks for the feedback. It helps me learn what people feel about various things. I saw the feedback and tried to modify the post to cater to your desires and recommendations. ", "votes": 2, "replies": [ { "id": 979638, "postDate": "2020-08-21T02:46:55.843Z", "content": "I like clear on the thing, I neither downvote your post nor upvote because I didn't find it worth to do that in both. The comment I posted by mentioned similar topic is simply let the people know and to stop double standard manner (upvote that one, but downvote this one). The competition is (most of the part) merely based on luck.
\nIMO, the solution from the top 3 teams is not surprising, not enough novelty there to be there, yet they placed. There are many top kaggler (i.e. 1st Comp. GM) couldn't achieve a single medal. What to blame, system! However, If I were you, I would ignore the hater.
", "rawMarkdown": "I like clear on the thing, I neither downvote your post nor upvote because I didn't find it worth to do that in both. The comment I posted by mentioned similar topic is simply let the people know and to stop double standard manner (upvote that one, but downvote this one). The competition is (most of the part) merely based on luck. \n\nIMO, the solution from the top 3 teams is not surprising, not enough novelty there to be there, yet they placed. There are many top kaggler (i.e. 1st Comp. GM) couldn't achieve a single medal. What to blame, system! However, If I were you, I would ignore the hater.", "votes": 1 } ] }, { "id": 978774, "postDate": "2020-08-20T11:32:32.130Z", "content": "The same type of post, but different feedback. The main difference I see is how one describe the things in his post.
\nI got a medal without doing any work
", "rawMarkdown": "The same type of post, but different feedback. The main difference I see is how one describe the things in his post. \n\n[I got a medal without doing any work](https://www.kaggle.com/c/bengaliai-cv19/discussion/136795)\n\n", "votes": 2 }, { "id": 975149, "postDate": "2020-08-18T07:12:13.957Z", "content": "@bopengiowa congrats for lucky silver ! can you tell how did you blended them to get this surprising results?
", "rawMarkdown": "@bopengiowa congrats for lucky silver ! can you tell how did you blended them to get this surprising results?", "votes": 2, "replies": [ { "id": 975179, "postDate": "2020-08-18T07:26:00.753Z", "rawMarkdown": "", "votes": -4, "isDeleted": true }, { "id": 975703, "postDate": "2020-08-18T12:32:43.617Z", "content": "shocked shocked shocked😲
", "rawMarkdown": "shocked shocked shocked😲 ", "votes": 1 } ] }, { "id": 975692, "postDate": "2020-08-18T12:27:07.083Z", "content": "I think this is a wonderful example of why people shouldn't worry so much about the cheating/medal sharing going on. Kaggle is a place to learn, with how unstable so many competitions are, the medals hold very little value anyway.
\ngrats on your medals nevertheless:)
", "rawMarkdown": "I think this is a wonderful example of why people shouldn't worry so much about the cheating/medal sharing going on. Kaggle is a place to learn, with how unstable so many competitions are, the medals hold very little value anyway.\n\ngrats on your medals nevertheless:)", "votes": -1 }, { "id": 975851, "postDate": "2020-08-18T13:50:34.930Z", "rawMarkdown": "", "isDeleted": true }, { "id": 975850, "postDate": "2020-08-18T13:50:34.567Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 975029, "author_name": "Nirjhar Roy", "author_url": "", "post_date": "2020-08-18T05:52:17.967000", "content": "First of all congrats on your high silver medal . Good for you . I have not downvoted you as I didn't want to bring sadness in your joy .
\nHowever there are few things that I do find a bit off :
\na) I have got a medal once just by tweaking a public model and submitting few hours before competition end . Was that illegal ? No . But can I promote that behaviour ? No . Because , I will then be misguiding a lot of people who probably would look upto kaggle community to learn . It was mere luck , there was no fun , I didn't learn anything .
\nb) Is ensembling bad ? No . But along with your post if you would have put the factors if any to chose model for ensemble , method of ensemble then probably people would also know that what are the tricks of the trade rather than probably blindly following a practice .
I think probably few people found these small points worth to downvote .
", "votes": 10, "replies": [] }, { "id": 975803, "author_name": "Optimo", "author_url": "", "post_date": "2020-08-18T13:26:14.140000", "content": "First good for you, second please be careful about how you phrase this: I did nothing. Because some us did work hard to come out with our solution, which may not be as good as blindly blending 20 public kernels but which has the taste of sweat, authenticity and hard work. To be honest I don't even understand why someone would spend time blending 20 public kernels but that's your problem not mine.
On a more interesting note, what happens if one of the public notebooks does not follow the competition rules? People should not be able to use it, but how would Kaggle know? This is an interesting question in my opinion and your example shows that public sharing can become an issue even in top 1-2% of Kaggle Competitions.
\nAs an example : Imagine someone hand labeled their test solution and called it a blend of blend. Then what happens to people who use it? Everyone is disqualified? How do you spot them?
", "votes": 7, "replies": [] }, { "id": 975077, "author_name": "Dieter", "author_url": "", "post_date": "2020-08-18T06:15:50.060000", "content": "Your arrogant bragging with getting a silver medal by doing nothing is a slap in the face for everybody that worked hard.
", "votes": 8, "replies": [ { "id": 975098, "author_name": "", "author_url": "", "post_date": "2020-08-18T06:37:52.787000", "content": "", "votes": -3, "replies": [] } ] }, { "id": 974651, "author_name": "Helen", "author_url": "", "post_date": "2020-08-18T02:17:31.940000", "content": "Congrats! So what kind of blending method did you use? I find postprocessing is quite important as well.
", "votes": 3, "replies": [] }, { "id": 974622, "author_name": "Sidney Ng", "author_url": "", "post_date": "2020-08-18T02:03:08.043000", "content": "Thanks for your honesty !
\nI did all the work by myself (refuse to use public kernels) --> Look where I got 8-(
", "votes": 3, "replies": [ { "id": 974637, "author_name": "Utkarsh Chandra Srivastava", "author_url": "", "post_date": "2020-08-18T02:09:15.377000", "content": "It's fine its much fun to work on our own cv then to submit solutions which you don't even understand
", "votes": 1, "replies": [] } ] }, { "id": 974629, "author_name": "Aman Arora", "author_url": "", "post_date": "2020-08-18T02:05:47.163000", "content": "Thanks for your honesty - appreciate it. :)
", "votes": 4, "replies": [] }, { "id": 974626, "author_name": "Bo Peng", "author_url": "", "post_date": "2020-08-18T02:04:52.680000", "content": "I'm just sharing what I did. I didn't have much time for Kaggle, saw this competition is very unstable, and submitted top ~20 blend. And got 34rd. I'm just sharing it encase someone finds interesting.
", "votes": 3, "replies": [ { "id": 974631, "author_name": "FGPC", "author_url": "", "post_date": "2020-08-18T02:06:20.010000", "content": "It means that you have great skills in model blending! :D. Thank you for your honesty!
", "votes": 1, "replies": [] } ] }, { "id": 974627, "author_name": "Khoi Nguyen", "author_url": "", "post_date": "2020-08-18T02:05:21.093000", "content": "Why the downvotes? Anyone cares to elaborate?
", "votes": 2, "replies": [] }, { "id": 975719, "author_name": "Aman Arora", "author_url": "", "post_date": "2020-08-18T12:44:18.033000", "content": "So you’re saying if I were to do this now - take 20 top public kernels and do simple average, I’d get rank 43 score? :)
", "votes": 1, "replies": [] }, { "id": 975085, "author_name": "Vadim Timakin", "author_url": "", "post_date": "2020-08-18T06:20:19.073000", "content": "M5 Deja vu. It's the fault of hosts, not yours, you shouldn't be ashamed of that.
", "votes": 1, "replies": [] }, { "id": 974898, "author_name": "Karan", "author_url": "", "post_date": "2020-08-18T04:46:01.750000", "content": "Congrats for winning a silver medal.
", "votes": 1, "replies": [] }, { "id": 974800, "author_name": "Jose", "author_url": "", "post_date": "2020-08-18T03:58:06.317000", "content": "Some of the high scoring public notebooks had good private scores like public 0.9605 and private 0.9446 which could have got you in top 40. On the other hand, some were really bad like public 0.9619 to private 0.9128. So yeah, luck is indeed a factor in the private leaderboard.😄
\nCongrats on your medal 💯
Appreciate your honesty @bopengiowa .. Congrats on your silver..
", "votes": 1, "replies": [] }, { "id": 974655, "author_name": "katsu1110", "author_url": "", "post_date": "2020-08-18T02:19:19.743000", "content": "LOL
\nWould you mind sharing how you blended those models? Any special blending techniques??
", "votes": 1, "replies": [] }, { "id": 974590, "author_name": "FGPC", "author_url": "", "post_date": "2020-08-18T01:48:58.457000", "content": "Interesting! LOL
", "votes": 1, "replies": [] }, { "id": 979184, "author_name": "Bo Peng", "author_url": "", "post_date": "2020-08-20T17:19:40.733000", "content": "Thanks for the feedback. It helps me learn what people feel about various things. I saw the feedback and tried to modify the post to cater to your desires and recommendations.
", "votes": 2, "replies": [ { "id": 979638, "author_name": "Innat", "author_url": "", "post_date": "2020-08-21T02:46:55.843000", "content": "I like clear on the thing, I neither downvote your post nor upvote because I didn't find it worth to do that in both. The comment I posted by mentioned similar topic is simply let the people know and to stop double standard manner (upvote that one, but downvote this one). The competition is (most of the part) merely based on luck.
\nIMO, the solution from the top 3 teams is not surprising, not enough novelty there to be there, yet they placed. There are many top kaggler (i.e. 1st Comp. GM) couldn't achieve a single medal. What to blame, system! However, If I were you, I would ignore the hater.
", "votes": 1, "replies": [] } ] }, { "id": 978774, "author_name": "Innat", "author_url": "", "post_date": "2020-08-20T11:32:32.130000", "content": "The same type of post, but different feedback. The main difference I see is how one describe the things in his post.
\nI got a medal without doing any work
", "votes": 2, "replies": [] }, { "id": 975149, "author_name": "Pawan KS", "author_url": "", "post_date": "2020-08-18T07:12:13.957000", "content": "@bopengiowa congrats for lucky silver ! can you tell how did you blended them to get this surprising results?
", "votes": 2, "replies": [ { "id": 975179, "author_name": "", "author_url": "", "post_date": "2020-08-18T07:26:00.753000", "content": "", "votes": -4, "replies": [] }, { "id": 975703, "author_name": "Salaryman", "author_url": "", "post_date": "2020-08-18T12:32:43.617000", "content": "shocked shocked shocked😲
", "votes": 1, "replies": [] } ] }, { "id": 975692, "author_name": "Proletheus", "author_url": "", "post_date": "2020-08-18T12:27:07.083000", "content": "I think this is a wonderful example of why people shouldn't worry so much about the cheating/medal sharing going on. Kaggle is a place to learn, with how unstable so many competitions are, the medals hold very little value anyway.
\ngrats on your medals nevertheless:)
", "votes": -1, "replies": [] }, { "id": 975851, "author_name": "", "author_url": "", "post_date": "2020-08-18T13:50:34.930000", "content": "", "votes": 0, "replies": [] }, { "id": 975850, "author_name": "", "author_url": "", "post_date": "2020-08-18T13:50:34.567000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "975029": "First of all congrats on your high silver medal . Good for you . I have not downvoted you as I didn't want to bring sadness in your joy . \n\nHowever there are few things that I do find a bit off :\na) I have got a medal once just by tweaking a public model and submitting few hours before competition end . Was that illegal ? No . But can I promote that behaviour ? No . Because , I will then be misguiding a lot of people who probably would look upto kaggle community to learn . It was mere luck , there was no fun , I didn't learn anything . \nb) Is ensembling bad ? No . But along with your post if you would have put the factors if any to chose model for ensemble , method of ensemble then probably people would also know that what are the tricks of the trade rather than probably blindly following a practice . \n\nI think probably few people found these small points worth to downvote . ", "974586": "I got some helpful feedback so I'm editing this post. Thanks for that. Basically I didn't have time for this competition, saw it was very unstable, and submitted two public kernel ensembles. \n\nI found sometimes in these very unstable competitions, there are alot of public overfitting that may not perform well in private LB, and sometimes a heavily ensembled solution ends up being stable enough to weather the shakeup. The submissions were one ensemble of the top 9 public lb kernels and the other one about the top ~20 public kernels. The top ~20 lb kernels blend scored 32nd place. It just used straight average. \n\nI did this before for the ASHRAE competition and I silvered that one as well. That ASHRAE competition was also unstable with crazy public overfitting if I remember correctly.\n\nI originally posted encase someone found it interesting. \n\nThis post has been edited.", "975803": "First good for you, second please be careful about how you phrase this: `I did nothing`. Because some us did work hard to come out with our solution, which may not be as good as blindly blending 20 public kernels but which has the taste of sweat, authenticity and hard work. To be honest I don't even understand why someone would spend time blending 20 public kernels but that's your problem not mine.\n\nOn a more interesting note, what happens if one of the public notebooks does not follow the competition rules? People should not be able to use it, but how would Kaggle know? This is an interesting question in my opinion and your example shows that public sharing can become an issue even in top 1-2% of Kaggle Competitions.\n\nAs an example : Imagine someone hand labeled their test solution and called it a blend of blend. Then what happens to people who use it? Everyone is disqualified? How do you spot them?\n\n\n", "975077": "Your arrogant bragging with getting a silver medal by doing nothing is a slap in the face for everybody that worked hard. ", "974651": "Congrats! So what kind of blending method did you use? I find postprocessing is quite important as well.", "974622": "Thanks for your honesty !\n\nI did all the work by myself (refuse to use public kernels) --> Look where I got 8-(", "974629": "Thanks for your honesty - appreciate it. :) ", "974626": "I'm just sharing what I did. I didn't have much time for Kaggle, saw this competition is very unstable, and submitted top ~20 blend. And got 34rd. I'm just sharing it encase someone finds interesting. ", "974627": "Why the downvotes? Anyone cares to elaborate?", "975719": "So you’re saying if I were to do this now - take 20 top public kernels and do simple average, I’d get rank 43 score? :)", "975085": "M5 Deja vu. It's the fault of hosts, not yours, you shouldn't be ashamed of that.", "974898": "Congrats for winning a silver medal.", "974800": "Some of the high scoring public notebooks had good private scores like public 0.9605 and private 0.9446 which could have got you in top 40. On the other hand, some were really bad like public 0.9619 to private 0.9128. So yeah, luck is indeed a factor in the private leaderboard.😄\nCongrats on your medal 💯", "974756": "Appreciate your honesty @bopengiowa .. Congrats on your silver.. ", "974655": "LOL \n\nWould you mind sharing how you blended those models? Any special blending techniques??", "974590": "Interesting! LOL", "979184": "Thanks for the feedback. It helps me learn what people feel about various things. I saw the feedback and tried to modify the post to cater to your desires and recommendations. ", "978774": "The same type of post, but different feedback. The main difference I see is how one describe the things in his post. \n\n[I got a medal without doing any work](https://www.kaggle.com/c/bengaliai-cv19/discussion/136795)\n\n", "975149": "@bopengiowa congrats for lucky silver ! can you tell how did you blended them to get this surprising results?", "975692": "I think this is a wonderful example of why people shouldn't worry so much about the cheating/medal sharing going on. Kaggle is a place to learn, with how unstable so many competitions are, the medals hold very little value anyway.\n\ngrats on your medals nevertheless:)", "975851": "", "975850": "" } }