![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1885266%2F8e9ed1cc864be12d549fdfc16845126e%2FScreen%20Shot%202020-08-20%20at%203.46.33%20PM.png?generation=1597952872763129&alt=media\")
Just revisited the Private LB and I now seem to have a Bronze!!! lol
\nThough I had 30 better scores, probably 5 of which would have given me silver. I'll take Bronze though ;)
\nThis seems to be the hardest part of Kaggle. Deciding what to submit finally.
\nEDIT: Sorry got excited and forgot about the resulting shrinkage of the medal zone (10%)
\nI'll be quiet until it's finalized lol
Just revisited the Private LB and I now seem to have a Bronze!!! lol
\nThough I had 30 better scores, probably 5 of which would have given me silver. I'll take Bronze though ;)
\nThis seems to be the hardest part of Kaggle. Deciding what to submit finally.
\nEDIT: Sorry got excited and forgot about the resulting shrinkage of the medal zone (10%)
\nI'll be quiet until it's finalized lol
It's official!
\n
Thanks for all the interest and support.
\nI have thought about \"what went wrong\" as like many of you, I had a far better submission not submitted for final assessment. I had a plan originally, largely based around Chris' multiple resolutions ideas. It was working well for me, but then I let it slide from 3 sizes to 2 with not much emphasis on the second. It was one of those earlier solutions that would have been up in the top 20 or 30 (I realise that many other people have the same \"if only\" story).
\nSo I have to tread more carefully and more cautiously evaluate \"good ideas\" from the discussion boards, even if they do make my public LB score higher. Doesn't mean I have to be inflexible but just step back a bit and consider the consequences more.
\nthanks to all again
", "rawMarkdown": "It's official!\n\n\n\nThanks for all the interest and support.\n\nI have thought about \"what went wrong\" as like many of you, I had a far better submission not submitted for final assessment. I had a plan originally, largely based around Chris' multiple resolutions ideas. It was working well for me, but then I let it slide from 3 sizes to 2 with not much emphasis on the second. It was one of those earlier solutions that would have been up in the top 20 or 30 (I realise that many other people have the same \"if only\" story).\n\nSo I have to tread more carefully and more cautiously evaluate \"good ideas\" from the discussion boards, even if they do make my public LB score higher. Doesn't mean I have to be inflexible but just step back a bit and consider the consequences more.\n\nthanks to all again", "votes": 11, "replies": [ { "id": 979555, "postDate": "2020-08-21T00:06:15.007Z", "content": "Congratulations @mutantspore! Well deserved! I learned from your contribution to discussions and also from your notebook in the flowers competition since I had no previous experience woking with TPU!
", "rawMarkdown": "Congratulations @mutantspore! Well deserved! I learned from your contribution to discussions and also from your notebook in the flowers competition since I had no previous experience woking with TPU!", "votes": 2 }, { "id": 979561, "postDate": "2020-08-21T00:19:22.977Z", "content": "Thanks Amin.
\nCongrats on your first medal too.
Congrats on bronze Bruce. Good job. Next comp you'll get silver and after that you'll get gold !!
", "rawMarkdown": "Congrats on bronze Bruce. Good job. Next comp you'll get silver and after that you'll get gold !!", "votes": 6 }, { "id": 979607, "postDate": "2020-08-21T01:55:19.177Z", "content": "Comgratulation! You did a good job! You deserve it!
", "rawMarkdown": "Comgratulation! You did a good job! You deserve it!", "votes": 1 }, { "id": 977188, "postDate": "2020-08-19T10:41:11.277Z", "content": "Congrats!
", "rawMarkdown": "Congrats!", "votes": 1 }, { "id": 976523, "postDate": "2020-08-18T23:33:36.067Z", "content": "Hope it will continue) from 410 to 358! Your result is great, nevertheless of medals, well done!
", "rawMarkdown": "Hope it will continue) from 410 to 358! Your result is great, nevertheless of medals, well done!", "votes": 1, "replies": [ { "id": 976534, "postDate": "2020-08-18T23:47:48.910Z", "content": "Thanks Iurii. Wondering in which competition I should try again.
", "rawMarkdown": "Thanks Iurii. Wondering in which competition I should try again." }, { "id": 976542, "postDate": "2020-08-18T23:58:42.067Z", "content": "@mutantspore Bruce, you did great. Don't worry about the medals.
\nI read your analysis of AgentAugers' notebook that you posted in this comments months ago and read everything you wrote carefully. In every one of my models I used #img = tf.image.random_hue(img, 0.01) because you said so. I also used your suggestions of epochs and backbones to guide my early experiments.
You are a great data scientist who conducts careful well thought out experiments. You were high on the leaderboard for most of the competition because of your good intuition. I'm looking forward to reading more of your comments in future competitions. And I know that you will get a medal in your next comp if you don't get one in this comp.
", "rawMarkdown": "@mutantspore Bruce, you did great. Don't worry about the medals. \n\nI read your analysis of AgentAugers' notebook that you posted in this comments months ago and read everything you wrote carefully. In every one of my models I used `#img = tf.image.random_hue(img, 0.01)` because you said so. I also used your suggestions of epochs and backbones to guide my early experiments.\n\nYou are a great data scientist who conducts careful well thought out experiments. You were high on the leaderboard for most of the competition because of your good intuition. I'm looking forward to reading more of your comments in future competitions. And I know that you will get a medal in your next comp if you don't get one in this comp.", "votes": 6 } ] }, { "id": 976274, "postDate": "2020-08-18T18:54:37.850Z", "content": "Oops Bruce you may still be out of Bronze range (cut off moved just a couple of places above your score) when more participants are removed, because of 10% cutoff (you are really at the edge of this limit) i.e. fewer participants means fewer medals ..
", "rawMarkdown": "Oops Bruce you may still be out of Bronze range (cut off moved just a couple of places above your score) when more participants are removed, because of 10% cutoff (you are really at the edge of this limit) i.e. fewer participants means fewer medals ..", "votes": 1, "replies": [ { "id": 976278, "postDate": "2020-08-18T18:58:14.280Z", "content": "Yep… Bronze has slipped out of my fingers… sorry for getting too excited too early.
", "rawMarkdown": "Yep... Bronze has slipped out of my fingers... sorry for getting too excited too early." } ] }, { "id": 976272, "postDate": "2020-08-18T18:53:04.447Z", "content": "Congrats Bruce on medal.
\nChoosing the final 3 was the toughest part for me too - especially ignoring the high ranking public blends and having faith in my local cross-validated scores.
Good luck for the future too.
", "rawMarkdown": "Congrats Bruce on medal.\nChoosing the final 3 was the toughest part for me too - especially ignoring the high ranking public blends and having faith in my local cross-validated scores.\n\nGood luck for the future too.", "votes": 1 }, { "id": 976262, "postDate": "2020-08-18T18:47:45.517Z", "content": "congrats!
", "rawMarkdown": "congrats! ", "votes": 1 }, { "id": 976241, "postDate": "2020-08-18T18:31:37.077Z", "content": "@mutantspore absolutely deciding what to submit finally is the hardest part.
\nlast competition i fell 450 places (and lost silver medal) because i did not pick the right submission :(
\nthis competition i moved up 677 places. I spent some time on picking the right submission.
\nit is by no means perfect but at-least now i had some heuristics i am happy with. I will make a discussion topic and share the link here.
Congrats Bruce, and sorry that you missed selecting a Silver submission …
\nYes, I confirm that my position moved up about 40 places as well!
", "rawMarkdown": "Congrats Bruce, and sorry that you missed selecting a Silver submission ...\n\nYes, I confirm that my position moved up about 40 places as well!", "votes": 1 }, { "id": 976233, "postDate": "2020-08-18T18:27:33.403Z", "content": "Just checked private LB and i barely earned bronze medal. But total teams (3437) is different with lowest \"#\" (3331).
", "rawMarkdown": "Just checked private LB and i barely earned bronze medal. But total teams (3437) is different with lowest \"#\" (3331).", "votes": 1 }, { "id": 976213, "postDate": "2020-08-18T18:06:28.903Z", "content": "I just paste here code from the same discussion:
\nYes, it is. I've asked hosts about this here. And it seems like they got it. However, it's not the end yet. You can easily make sure that there are still a lot of empty accounts in the medal zone.
", "rawMarkdown": "I just paste here code from the same discussion:\n\nYes, it is. I've asked hosts about this [here](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175521). And it seems like they got it. However, it's not the end yet. You can easily make sure that there are still a lot of empty accounts in the medal zone.", "votes": 1 }, { "id": 979363, "postDate": "2020-08-20T19:28:03.560Z", "content": "Congratulations @mutantspore you are back in Bronze now at #330 after more LB cleanup!
", "rawMarkdown": "Congratulations @mutantspore **you are back** in Bronze now at #330 after more LB cleanup!", "votes": 2, "replies": [ { "id": 979372, "postDate": "2020-08-20T19:32:32.897Z", "content": "WooooHooooo!!!
\nand it's finalised to.. set in stone… done
\nYES!
\nthanks for letting me know Sirish.
", "rawMarkdown": "***WooooHooooo!!!***\n\nand it's finalised to.. set in stone... done\n\n**YES!**\n\nthanks for letting me know Sirish.\n", "votes": 2 }, { "id": 979374, "postDate": "2020-08-20T19:35:53.017Z", "content": "Is it really finalised?
\nI see the same number of teams and don't see the medal in my profile.
\nIt seems like there's just something wrong with the banner about \"preliminary standings\".
well … it's not \"set in stone\" until you see counter \"1\" under Bronze on your home page!
\nonce again, contain your excitement and patiently wait a few hours / days longer 😂
See here for more updates
", "rawMarkdown": "well ... it's not \"set in stone\" until you see counter \"1\" under Bronze on your home page!\nonce again, contain your excitement and patiently wait a few hours / days longer 😂\n\nSee [here](https://www.kaggle.com/general/176169) for more updates" }, { "id": 979380, "postDate": "2020-08-20T19:38:52.017Z", "content": "yes… I don't see an awarded Bronze but I do see this and no Orange disclaimer.
\nThe private leaderboard is calculated with approximately 70% of the test data.\nThis competition has completed. This leaderboard reflects the final standings.\nok ok … quiet again… waiting..
\nI guess if the number of teams hasn't changed, I could have the same thing happen again when it's adjusted… Just outside the 10%.
", "rawMarkdown": "yes... I don't see an awarded Bronze but I do see this and no Orange disclaimer.\n\n```\nThe private leaderboard is calculated with approximately 70% of the test data.\nThis competition has completed. This leaderboard reflects the final standings.\n\n```\nok ok ... quiet again... waiting..\n\nI guess if the number of teams hasn't changed, I could have the same thing happen again when it's adjusted... Just outside the 10%. \n", "votes": 1 }, { "id": 979388, "postDate": "2020-08-20T19:44:06.530Z", "content": "You are right! It's final now and I just got my Bronze count updated!
\nand I became a ranked Competitions Expert (#1803 is not too bad to begin with 😏)
I got my first medal too and I jumped 2 places from 328 to 326 now.
", "rawMarkdown": "I got my first medal too and I jumped 2 places from 328 to 326 now.", "votes": 2 } ] }, { "id": 976544, "postDate": "2020-08-19T00:02:08.990Z", "content": "In @agentauers awesome notebook, i saw your reply and implemented it like you said
\nyou replied that i remember you got a score around 0.9588 using noisy student with the combination of 256, 384, 512 sizes efficientnet and adjusting crop size with metadata + delete #tf.image.randome_hue i thought it makes sense) and at the same time i could get 0.9600 Public Leaderboard score and accordingly Private score now i checked just now 0.9376
based on your reply, i made around 37 models composed of noisy student from 384 size to 1024 size 5 fold stratifed
\nwith just 37 models , my CV was approximately 0.952 accordingly my private leaderboard score clustered around 0.942~0.945 (i checked just now!)
it was really helpful. you are doing very well!
\nKeep trying!
", "rawMarkdown": "In @agentauers awesome notebook, i saw your reply and implemented it like you said\nyou replied that i remember you got a score around 0.9588 using noisy student with the combination of 256, 384, 512 sizes efficientnet and adjusting crop size with metadata + delete #tf.image.randome_hue i thought it makes sense) and at the same time i could get 0.9600 Public Leaderboard score and accordingly Private score now i checked just now 0.9376\n\nbased on your reply, i made around 37 models composed of noisy student from 384 size to 1024 size 5 fold stratifed\nwith just 37 models , my CV was approximately 0.952 accordingly my private leaderboard score clustered around 0.942~0.945 (i checked just now!)\n\nit was really helpful. you are doing very well! \n\nKeep trying!\n", "votes": 2 }, { "id": 976259, "postDate": "2020-08-18T18:44:48.793Z", "content": "Congrats Bruce, that's a deserved medal, happy for you! I just moved up to edge of solo gold medal, can't we find 2-3 more cheaters 😁
\n(Jokes aside congrats to all on top ranking)
", "rawMarkdown": "Congrats Bruce, that's a deserved medal, happy for you! I just moved up to edge of solo gold medal, can't we find 2-3 more cheaters 😁\n\n(Jokes aside congrats to all on top ranking)", "votes": 2, "replies": [ { "id": 976265, "postDate": "2020-08-18T18:48:54.030Z", "content": "lol my \"if only\" score was 0.9456 which is right about where you are now 😄
", "rawMarkdown": "lol my \"if only\" score was **0.9456** which is right about where you are now 😄", "votes": 1 } ] }, { "id": 980267, "postDate": "2020-08-21T12:57:00.597Z", "rawMarkdown": "", "votes": -1, "isDeleted": true } ], "comments": [ { "id": 979401, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-20T19:58:57.433000", "content": "It's official!
\n
Thanks for all the interest and support.
\nI have thought about \"what went wrong\" as like many of you, I had a far better submission not submitted for final assessment. I had a plan originally, largely based around Chris' multiple resolutions ideas. It was working well for me, but then I let it slide from 3 sizes to 2 with not much emphasis on the second. It was one of those earlier solutions that would have been up in the top 20 or 30 (I realise that many other people have the same \"if only\" story).
\nSo I have to tread more carefully and more cautiously evaluate \"good ideas\" from the discussion boards, even if they do make my public LB score higher. Doesn't mean I have to be inflexible but just step back a bit and consider the consequences more.
\nthanks to all again
", "votes": 11, "replies": [ { "id": 979555, "author_name": "Amin", "author_url": "", "post_date": "2020-08-21T00:06:15.007000", "content": "Congratulations @mutantspore! Well deserved! I learned from your contribution to discussions and also from your notebook in the flowers competition since I had no previous experience woking with TPU!
", "votes": 2, "replies": [] }, { "id": 979561, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-21T00:19:22.977000", "content": "Thanks Amin.
\nCongrats on your first medal too.
Congrats on bronze Bruce. Good job. Next comp you'll get silver and after that you'll get gold !!
", "votes": 6, "replies": [] }, { "id": 979607, "author_name": "kaggler", "author_url": "", "post_date": "2020-08-21T01:55:19.177000", "content": "Comgratulation! You did a good job! You deserve it!
", "votes": 1, "replies": [] }, { "id": 977188, "author_name": "Abishek Sudarshan", "author_url": "", "post_date": "2020-08-19T10:41:11.277000", "content": "Congrats!
", "votes": 1, "replies": [] }, { "id": 976523, "author_name": "Iurii Uspangaliev", "author_url": "", "post_date": "2020-08-18T23:33:36.067000", "content": "Hope it will continue) from 410 to 358! Your result is great, nevertheless of medals, well done!
", "votes": 1, "replies": [ { "id": 976534, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-18T23:47:48.910000", "content": "Thanks Iurii. Wondering in which competition I should try again.
", "votes": 0, "replies": [] }, { "id": 976542, "author_name": "Chris Deotte", "author_url": "", "post_date": "2020-08-18T23:58:42.067000", "content": "@mutantspore Bruce, you did great. Don't worry about the medals.
\nI read your analysis of AgentAugers' notebook that you posted in this comments months ago and read everything you wrote carefully. In every one of my models I used #img = tf.image.random_hue(img, 0.01) because you said so. I also used your suggestions of epochs and backbones to guide my early experiments.
You are a great data scientist who conducts careful well thought out experiments. You were high on the leaderboard for most of the competition because of your good intuition. I'm looking forward to reading more of your comments in future competitions. And I know that you will get a medal in your next comp if you don't get one in this comp.
", "votes": 6, "replies": [] } ] }, { "id": 976274, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2020-08-18T18:54:37.850000", "content": "Oops Bruce you may still be out of Bronze range (cut off moved just a couple of places above your score) when more participants are removed, because of 10% cutoff (you are really at the edge of this limit) i.e. fewer participants means fewer medals ..
", "votes": 1, "replies": [ { "id": 976278, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-18T18:58:14.280000", "content": "Yep… Bronze has slipped out of my fingers… sorry for getting too excited too early.
", "votes": 0, "replies": [] } ] }, { "id": 976272, "author_name": "VSR", "author_url": "", "post_date": "2020-08-18T18:53:04.447000", "content": "Congrats Bruce on medal.
\nChoosing the final 3 was the toughest part for me too - especially ignoring the high ranking public blends and having faith in my local cross-validated scores.
Good luck for the future too.
", "votes": 1, "replies": [] }, { "id": 976262, "author_name": "yimacs", "author_url": "", "post_date": "2020-08-18T18:47:45.517000", "content": "congrats!
", "votes": 1, "replies": [] }, { "id": 976241, "author_name": "Mehul Sampat", "author_url": "", "post_date": "2020-08-18T18:31:37.077000", "content": "@mutantspore absolutely deciding what to submit finally is the hardest part.
\nlast competition i fell 450 places (and lost silver medal) because i did not pick the right submission :(
\nthis competition i moved up 677 places. I spent some time on picking the right submission.
\nit is by no means perfect but at-least now i had some heuristics i am happy with. I will make a discussion topic and share the link here.
Congrats Bruce, and sorry that you missed selecting a Silver submission …
\nYes, I confirm that my position moved up about 40 places as well!
", "votes": 1, "replies": [] }, { "id": 976233, "author_name": "Sandy Khosasi", "author_url": "", "post_date": "2020-08-18T18:27:33.403000", "content": "Just checked private LB and i barely earned bronze medal. But total teams (3437) is different with lowest \"#\" (3331).
", "votes": 1, "replies": [] }, { "id": 976213, "author_name": "Vadim Timakin", "author_url": "", "post_date": "2020-08-18T18:06:28.903000", "content": "I just paste here code from the same discussion:
\nYes, it is. I've asked hosts about this here. And it seems like they got it. However, it's not the end yet. You can easily make sure that there are still a lot of empty accounts in the medal zone.
", "votes": 1, "replies": [] }, { "id": 979363, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2020-08-20T19:28:03.560000", "content": "Congratulations @mutantspore you are back in Bronze now at #330 after more LB cleanup!
", "votes": 2, "replies": [ { "id": 979372, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-20T19:32:32.897000", "content": "WooooHooooo!!!
\nand it's finalised to.. set in stone… done
\nYES!
\nthanks for letting me know Sirish.
", "votes": 2, "replies": [] }, { "id": 979374, "author_name": "Stanislav Blinov", "author_url": "", "post_date": "2020-08-20T19:35:53.017000", "content": "Is it really finalised?
\nI see the same number of teams and don't see the medal in my profile.
\nIt seems like there's just something wrong with the banner about \"preliminary standings\".
well … it's not \"set in stone\" until you see counter \"1\" under Bronze on your home page!
\nonce again, contain your excitement and patiently wait a few hours / days longer 😂
See here for more updates
", "votes": 0, "replies": [] }, { "id": 979380, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-20T19:38:52.017000", "content": "yes… I don't see an awarded Bronze but I do see this and no Orange disclaimer.
\nThe private leaderboard is calculated with approximately 70% of the test data.\nThis competition has completed. This leaderboard reflects the final standings.\nok ok … quiet again… waiting..
\nI guess if the number of teams hasn't changed, I could have the same thing happen again when it's adjusted… Just outside the 10%.
", "votes": 1, "replies": [] }, { "id": 979388, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2020-08-20T19:44:06.530000", "content": "You are right! It's final now and I just got my Bronze count updated!
\nand I became a ranked Competitions Expert (#1803 is not too bad to begin with 😏)
I got my first medal too and I jumped 2 places from 328 to 326 now.
", "votes": 2, "replies": [] } ] }, { "id": 976544, "author_name": "kaggler", "author_url": "", "post_date": "2020-08-19T00:02:08.990000", "content": "In @agentauers awesome notebook, i saw your reply and implemented it like you said
\nyou replied that i remember you got a score around 0.9588 using noisy student with the combination of 256, 384, 512 sizes efficientnet and adjusting crop size with metadata + delete #tf.image.randome_hue i thought it makes sense) and at the same time i could get 0.9600 Public Leaderboard score and accordingly Private score now i checked just now 0.9376
based on your reply, i made around 37 models composed of noisy student from 384 size to 1024 size 5 fold stratifed
\nwith just 37 models , my CV was approximately 0.952 accordingly my private leaderboard score clustered around 0.942~0.945 (i checked just now!)
it was really helpful. you are doing very well!
\nKeep trying!
", "votes": 2, "replies": [] }, { "id": 976259, "author_name": "Ertuğrul Demir", "author_url": "", "post_date": "2020-08-18T18:44:48.793000", "content": "Congrats Bruce, that's a deserved medal, happy for you! I just moved up to edge of solo gold medal, can't we find 2-3 more cheaters 😁
\n(Jokes aside congrats to all on top ranking)
", "votes": 2, "replies": [ { "id": 976265, "author_name": "Bruce Young", "author_url": "", "post_date": "2020-08-18T18:48:54.030000", "content": "lol my \"if only\" score was 0.9456 which is right about where you are now 😄
", "votes": 1, "replies": [] } ] }, { "id": 980267, "author_name": "", "author_url": "", "post_date": "2020-08-21T12:57:00.597000", "content": "", "votes": -1, "replies": [] } ], "raw_markdown_by_id": { "976207": "Just revisited the Private LB and I now seem to have a Bronze!!! lol\n\nThough I had 30 better scores, probably 5 of which would have given me silver. I'll take Bronze though ;)\n\nThis seems to be the hardest part of Kaggle. Deciding what to submit finally. \n\nEDIT: Sorry got excited and forgot about the resulting shrinkage of the medal zone (10%)\nI'll be quiet until it's finalized lol", "979401": "It's official!\n\n\n\nThanks for all the interest and support.\n\nI have thought about \"what went wrong\" as like many of you, I had a far better submission not submitted for final assessment. I had a plan originally, largely based around Chris' multiple resolutions ideas. It was working well for me, but then I let it slide from 3 sizes to 2 with not much emphasis on the second. It was one of those earlier solutions that would have been up in the top 20 or 30 (I realise that many other people have the same \"if only\" story).\n\nSo I have to tread more carefully and more cautiously evaluate \"good ideas\" from the discussion boards, even if they do make my public LB score higher. Doesn't mean I have to be inflexible but just step back a bit and consider the consequences more.\n\nthanks to all again", "976227": "Congrats on bronze Bruce. Good job. Next comp you'll get silver and after that you'll get gold !!", "979607": "Comgratulation! You did a good job! You deserve it!", "977188": "Congrats!", "976523": "Hope it will continue) from 410 to 358! Your result is great, nevertheless of medals, well done!", "976274": "Oops Bruce you may still be out of Bronze range (cut off moved just a couple of places above your score) when more participants are removed, because of 10% cutoff (you are really at the edge of this limit) i.e. fewer participants means fewer medals ..", "976272": "Congrats Bruce on medal.\nChoosing the final 3 was the toughest part for me too - especially ignoring the high ranking public blends and having faith in my local cross-validated scores.\n\nGood luck for the future too.", "976262": "congrats! ", "976241": "@mutantspore absolutely deciding what to submit finally is the hardest part. \nlast competition i fell 450 places (and lost silver medal) because i did not pick the right submission :( \nthis competition i moved up 677 places. I spent some time on picking the right submission. \nit is by no means perfect but at-least now i had some heuristics i am happy with. I will make a discussion topic and share the link here. ", "976239": "Congrats Bruce, and sorry that you missed selecting a Silver submission ...\n\nYes, I confirm that my position moved up about 40 places as well!", "976233": "Just checked private LB and i barely earned bronze medal. But total teams (3437) is different with lowest \"#\" (3331).", "976213": "I just paste here code from the same discussion:\n\nYes, it is. I've asked hosts about this [here](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175521). And it seems like they got it. However, it's not the end yet. You can easily make sure that there are still a lot of empty accounts in the medal zone.", "979363": "Congratulations @mutantspore **you are back** in Bronze now at #330 after more LB cleanup!", "976544": "In @agentauers awesome notebook, i saw your reply and implemented it like you said\nyou replied that i remember you got a score around 0.9588 using noisy student with the combination of 256, 384, 512 sizes efficientnet and adjusting crop size with metadata + delete #tf.image.randome_hue i thought it makes sense) and at the same time i could get 0.9600 Public Leaderboard score and accordingly Private score now i checked just now 0.9376\n\nbased on your reply, i made around 37 models composed of noisy student from 384 size to 1024 size 5 fold stratifed\nwith just 37 models , my CV was approximately 0.952 accordingly my private leaderboard score clustered around 0.942~0.945 (i checked just now!)\n\nit was really helpful. you are doing very well! \n\nKeep trying!\n", "976259": "Congrats Bruce, that's a deserved medal, happy for you! I just moved up to edge of solo gold medal, can't we find 2-3 more cheaters 😁\n\n(Jokes aside congrats to all on top ranking)", "980267": "" } }