Ok, that was a huge shakeup. It's really disheartening to see most of the top teams who may have overfit public Lb a little bit or chose wrong submissions are down to 4 digit ranks. Meanwhile, there are some winning gold medals with only 1 submission, do you really think its fair?
\nI just want to express my gratitude towards those who shared brilliant ideas and worked hard for months but not able to succeed in private LB. Special mention to @carlossouza @sandorkonya @ChristianDenich @khoongweihao and many more…..
", "messages": [ { "id": "1040321", "postDate": "10/07/2020 04:50:33", "content": "Ok, that was a huge shakeup. It's really disheartening to see most of the top teams who may have overfit public Lb a little bit or chose wrong submissions are down to 4 digit ranks. Meanwhile, there are some winning gold medals with only 1 submission, do you really think its fair?
\nI just want to express my gratitude towards those who shared brilliant ideas and worked hard for months but not able to succeed in private LB. Special mention to @carlossouza @sandorkonya @ChristianDenich @khoongweihao and many more…..
", "rawMarkdown": "Ok, that was a huge shakeup. It's really disheartening to see most of the top teams who may have overfit public Lb a little bit or chose wrong submissions are down to 4 digit ranks. Meanwhile, there are some winning gold medals with only 1 submission, do you really think its fair? \n\nI just want to express my gratitude towards those who shared brilliant ideas and worked hard for months but not able to succeed in private LB. Special mention to @carlossouza @sandorkonya @ChristianDenich @khoongweihao and many more.....", "votes": null }, { "id": "1040355", "postDate": "10/07/2020 05:13:03", "content": "Seems like there was a huge difference between the public test set and the private one.
", "rawMarkdown": "Seems like there was a huge difference between the public test set and the private one.", "votes": null }, { "id": "1040471", "postDate": "10/07/2020 06:48:15", "content": "Yes, it is fair. Some people make their solutions public to farm medals, and some people run these solutions.
\nI'm pretty sure Alex Petrov just took one of those \"EfficientNet + Quantile Regression\" notebooks which got him gold medal with only 1 submission. I see no reason to complain because everyone got what they wanted to get -- folks with shared forks of \"EfficientNet + Quantile Regression\" got their notebooks medals, and some folks who ran it got their medals for this competition. I guess it's just not worth to complain about such stuff because it's much better to learn something on your way to pursue the goal of getting medals, than just farm it by bruteforcing the hyperparameters.
", "rawMarkdown": "Yes, it is fair. Some people make their solutions public to farm medals, and some people run these solutions.\n\nI'm pretty sure Alex Petrov just took one of those \"EfficientNet + Quantile Regression\" notebooks which got him gold medal with only 1 submission. I see no reason to complain because everyone got what they wanted to get -- folks with shared forks of \"EfficientNet + Quantile Regression\" got their notebooks medals, and some folks who ran it got their medals for this competition. I guess it's just not worth to complain about such stuff because it's much better to learn something on your way to pursue the goal of getting medals, than just farm it by bruteforcing the hyperparameters.", "votes": null }, { "id": "1040475", "postDate": "10/07/2020 06:51:15", "content": "Dear @nischaydnk, thank you for your cheering-up post!
\nThis time the work we put in and the result really are in no relation. Anyway, we learned a lot and will move on.
Thanks again for your kind words!
", "rawMarkdown": "Dear @nischaydnk, thank you for your cheering-up post!\nThis time the work we put in and the result really are in no relation. Anyway, we learned a lot and will move on.\n\nThanks again for your kind words!", "votes": null }, { "id": "1040565", "postDate": "10/07/2020 08:21:21", "content": "maaro mujhe xD
", "rawMarkdown": "maaro mujhe xD", "votes": null }, { "id": "1040580", "postDate": "10/07/2020 08:27:11", "content": "I really appreciate the work you guys had done ! i personally with my team had made more then 350 submission ,remember that this is just a begining ! many more medals to come !
", "rawMarkdown": "I really appreciate the work you guys had done ! i personally with my team had made more then 350 submission ,remember that this is just a begining ! many more medals to come !", "votes": null }, { "id": "1041257", "postDate": "10/07/2020 16:44:58", "content": "Learned a lot from your work!
\nIn my opinion, if \"contribution to this competition\" was quantified properly and the leaderboard was sorted by that ,you would win the gold medal without doubt.
Thank you for your hard work and effort, sincerely.
", "rawMarkdown": "Learned a lot from your work! \nIn my opinion, if \"contribution to this competition\" was quantified properly and the leaderboard was sorted by that ,you would win the gold medal without doubt. \n\nThank you for your hard work and effort, sincerely.", "votes": null }, { "id": "1044963", "postDate": "10/10/2020 08:52:27", "content": "It is not pleasant, but we need to take it with philosophy, at the end of the day we are here for learning new stuffs :)
", "rawMarkdown": "It is not pleasant, but we need to take it with philosophy, at the end of the day we are here for learning new stuffs :)", "votes": null } ], "comments": [ { "id": 1040355, "author_name": "imoore", "author_url": "", "post_date": "10/07/2020 05:13:03", "content": "Seems like there was a huge difference between the public test set and the private one.
", "votes": null, "replies": [] }, { "id": 1040471, "author_name": "benefactor", "author_url": "", "post_date": "10/07/2020 06:48:15", "content": "Yes, it is fair. Some people make their solutions public to farm medals, and some people run these solutions.
\nI'm pretty sure Alex Petrov just took one of those \"EfficientNet + Quantile Regression\" notebooks which got him gold medal with only 1 submission. I see no reason to complain because everyone got what they wanted to get -- folks with shared forks of \"EfficientNet + Quantile Regression\" got their notebooks medals, and some folks who ran it got their medals for this competition. I guess it's just not worth to complain about such stuff because it's much better to learn something on your way to pursue the goal of getting medals, than just farm it by bruteforcing the hyperparameters.
", "votes": null, "replies": [] }, { "id": 1040475, "author_name": "ChristianDenich", "author_url": "", "post_date": "10/07/2020 06:51:15", "content": "Dear @nischaydnk, thank you for your cheering-up post!
\nThis time the work we put in and the result really are in no relation. Anyway, we learned a lot and will move on.
Thanks again for your kind words!
", "votes": null, "replies": [ { "id": 1040580, "author_name": "trooperog", "author_url": "", "post_date": "10/07/2020 08:27:11", "content": "I really appreciate the work you guys had done ! i personally with my team had made more then 350 submission ,remember that this is just a begining ! many more medals to come !
", "votes": null, "replies": [] }, { "id": 1041257, "author_name": "rapisu0", "author_url": "", "post_date": "10/07/2020 16:44:58", "content": "Learned a lot from your work!
\nIn my opinion, if \"contribution to this competition\" was quantified properly and the leaderboard was sorted by that ,you would win the gold medal without doubt.
Thank you for your hard work and effort, sincerely.
", "votes": null, "replies": [] } ] }, { "id": 1040565, "author_name": "thakurudit", "author_url": "", "post_date": "10/07/2020 08:21:21", "content": "maaro mujhe xD
", "votes": null, "replies": [] }, { "id": 1044963, "author_name": "domizianostingi", "author_url": "", "post_date": "10/10/2020 08:52:27", "content": "It is not pleasant, but we need to take it with philosophy, at the end of the day we are here for learning new stuffs :)
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1040321": "Ok, that was a huge shakeup. It's really disheartening to see most of the top teams who may have overfit public Lb a little bit or chose wrong submissions are down to 4 digit ranks. Meanwhile, there are some winning gold medals with only 1 submission, do you really think its fair? \n\nI just want to express my gratitude towards those who shared brilliant ideas and worked hard for months but not able to succeed in private LB. Special mention to @carlossouza @sandorkonya @ChristianDenich @khoongweihao and many more.....", "1040355": "Seems like there was a huge difference between the public test set and the private one.", "1040471": "Yes, it is fair. Some people make their solutions public to farm medals, and some people run these solutions.\n\nI'm pretty sure Alex Petrov just took one of those \"EfficientNet + Quantile Regression\" notebooks which got him gold medal with only 1 submission. I see no reason to complain because everyone got what they wanted to get -- folks with shared forks of \"EfficientNet + Quantile Regression\" got their notebooks medals, and some folks who ran it got their medals for this competition. I guess it's just not worth to complain about such stuff because it's much better to learn something on your way to pursue the goal of getting medals, than just farm it by bruteforcing the hyperparameters.", "1040475": "Dear @nischaydnk, thank you for your cheering-up post!\nThis time the work we put in and the result really are in no relation. Anyway, we learned a lot and will move on.\n\nThanks again for your kind words!", "1040565": "maaro mujhe xD", "1040580": "I really appreciate the work you guys had done ! i personally with my team had made more then 350 submission ,remember that this is just a begining ! many more medals to come !", "1041257": "Learned a lot from your work! \nIn my opinion, if \"contribution to this competition\" was quantified properly and the leaderboard was sorted by that ,you would win the gold medal without doubt. \n\nThank you for your hard work and effort, sincerely.", "1044963": "It is not pleasant, but we need to take it with philosophy, at the end of the day we are here for learning new stuffs :)" }, "source": "meta" }