![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1341302%2Ff2bfd65d980ad56d0d55b84ff8715cfd%2Fbest_score.PNG?generation=1574122709281531&alt=media\")
Congratulations to winners and golds. Also, congratulation to everyone who learns a lot from this contest.
\n\nIn the last few days, I was getting funny results from different ensembles. Thus, I select low scored submissions (0.6732 and 0.6739) for finals. Guess what, the private LB came better for 0.6732 one. That's not the funny part. The most interesting part is in the following image.
\n\nMy best score in private LB is 0.66758. But for the same submission public LB was 0.66654. \nIt came from a single InceptionresnetV2 network 4-fold.
\n\n
Finally, I would like to give billions of thanks to @sgalib. Throughout this contest, I learn a lot from him. He encouraged me much to join in the kaggle competitions 2 years ago and taught me lots of ML concepts. I couldn't able to do this much without his cordial supports, lessons, and teaching. Personally, I am so grateful to him.
", "messages": [ { "id": 676075, "postDate": "2019-11-19T00:25:54.417Z", "content": "Congratulations to winners and golds. Also, congratulation to everyone who learns a lot from this contest.
\n\nIn the last few days, I was getting funny results from different ensembles. Thus, I select low scored submissions (0.6732 and 0.6739) for finals. Guess what, the private LB came better for 0.6732 one. That's not the funny part. The most interesting part is in the following image.
\n\nMy best score in private LB is 0.66758. But for the same submission public LB was 0.66654. \nIt came from a single InceptionresnetV2 network 4-fold.
\n\n
Finally, I would like to give billions of thanks to @sgalib. Throughout this contest, I learn a lot from him. He encouraged me much to join in the kaggle competitions 2 years ago and taught me lots of ML concepts. I couldn't able to do this much without his cordial supports, lessons, and teaching. Personally, I am so grateful to him.
", "rawMarkdown": "Congratulations to winners and golds. Also, congratulation to everyone who learns a lot from this contest. \n\nIn the last few days, I was getting funny results from different ensembles. Thus, I select low scored submissions (0.6732 and 0.6739) for finals. Guess what, the private LB came better for 0.6732 one. That's not the funny part. The most interesting part is in the following image. \n\n**My best score in private LB is 0.66758. But for the same submission public LB was 0.66654**. \nIt came from a single InceptionresnetV2 network 4-fold. \n\n\n\nFinally, I would like to give billions of thanks to @sgalib. Throughout this contest, I learn a lot from him. He encouraged me much to join in the kaggle competitions 2 years ago and taught me lots of ML concepts. I couldn't able to do this much without his cordial supports, lessons, and teaching. Personally, I am so grateful to him. \n", "votes": 15 }, { "id": 677808, "postDate": "2019-11-20T16:00:59.613Z", "content": "Wow that's great @mykttu . Looking forward to your solution.
", "rawMarkdown": "Wow that's great @mykttu . Looking forward to your solution. ", "votes": 1 }, { "id": 676086, "postDate": "2019-11-19T00:33:14.753Z", "content": "Nice. I couldn't fix my ensemble code but I still got bronze with only a single-fold efficientnetb2
", "rawMarkdown": "Nice. I couldn't fix my ensemble code but I still got bronze with only a single-fold efficientnetb2", "votes": 1, "replies": [ { "id": 676095, "postDate": "2019-11-19T00:43:27.810Z", "content": "Great. Congratulations. :)
", "rawMarkdown": "Great. Congratulations. :) " } ] }, { "id": 676080, "postDate": "2019-11-19T00:30:05.390Z", "content": "Wow, that's impressive. Will you share the write up about this single network? I look forward to see it :)
", "rawMarkdown": "Wow, that's impressive. Will you share the write up about this single network? I look forward to see it :)", "votes": 1, "replies": [ { "id": 676101, "postDate": "2019-11-19T00:48:07.880Z", "content": "Thanks. I will write up after a few days.
", "rawMarkdown": "Thanks. I will write up after a few days. ", "votes": 1 } ] }, { "id": 678344, "postDate": "2019-11-21T08:59:59.747Z", "content": "Congratulations !........Great work.
", "rawMarkdown": "Congratulations !........Great work." }, { "id": 676285, "postDate": "2019-11-19T04:21:12.333Z", "content": "Sad story shared: I used to have my best submission two weeks before the end.
", "rawMarkdown": "Sad story shared: I used to have my best submission two weeks before the end." }, { "id": 676255, "postDate": "2019-11-19T03:45:58.993Z", "content": "Interesting ✨ . Thanks for sharing, and congratulations! 🎉
", "rawMarkdown": "Interesting ✨ . Thanks for sharing, and congratulations! 🎉 " }, { "id": 676154, "postDate": "2019-11-19T01:56:58.247Z", "content": "Can you please share your code? I am really curious about lovasz loss working so well for this data.
", "rawMarkdown": "Can you please share your code? I am really curious about lovasz loss working so well for this data.", "replies": [ { "id": 676161, "postDate": "2019-11-19T02:06:11.790Z", "content": "I will share a writeup soon.
", "rawMarkdown": "I will share a writeup soon. " } ] }, { "id": 676152, "postDate": "2019-11-19T01:54:33.513Z", "content": "congratulation, where can we find @sgalib's lessons:)
", "rawMarkdown": "congratulation, where can we find @sgalib's lessons:)" }, { "id": 676084, "postDate": "2019-11-19T00:31:39.753Z", "content": "thanks for the share. is it because of lovaz loss?
", "rawMarkdown": "thanks for the share. is it because of lovaz loss?", "replies": [ { "id": 676091, "postDate": "2019-11-19T00:37:35.667Z", "content": "I think so.
", "rawMarkdown": "I think so. ", "votes": 1 }, { "id": 676096, "postDate": "2019-11-19T00:43:39.680Z", "content": "I think the train and test dataset were similar. So the model gave the best CV scored best in private LB.
", "rawMarkdown": "I think the train and test dataset were similar. So the model gave the best CV scored best in private LB. ", "votes": 1 } ] }, { "id": 676345, "postDate": "2019-11-19T05:21:48.913Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 677808, "author_name": "Manoj Prabhakar", "author_url": "", "post_date": "2019-11-20T16:00:59.613000", "content": "Wow that's great @mykttu . Looking forward to your solution.
", "votes": 1, "replies": [] }, { "id": 676086, "author_name": "c-weed", "author_url": "", "post_date": "2019-11-19T00:33:14.753000", "content": "Nice. I couldn't fix my ensemble code but I still got bronze with only a single-fold efficientnetb2
", "votes": 1, "replies": [ { "id": 676095, "author_name": "Md Yasin Kabir", "author_url": "", "post_date": "2019-11-19T00:43:27.810000", "content": "Great. Congratulations. :)
", "votes": 0, "replies": [] } ] }, { "id": 676080, "author_name": "Tsai29", "author_url": "", "post_date": "2019-11-19T00:30:05.390000", "content": "Wow, that's impressive. Will you share the write up about this single network? I look forward to see it :)
", "votes": 1, "replies": [ { "id": 676101, "author_name": "Md Yasin Kabir", "author_url": "", "post_date": "2019-11-19T00:48:07.880000", "content": "Thanks. I will write up after a few days.
", "votes": 1, "replies": [] } ] }, { "id": 678344, "author_name": "HAFIZ ALI E", "author_url": "", "post_date": "2019-11-21T08:59:59.747000", "content": "Congratulations !........Great work.
", "votes": 0, "replies": [] }, { "id": 676285, "author_name": "Endi Niu", "author_url": "", "post_date": "2019-11-19T04:21:12.333000", "content": "Sad story shared: I used to have my best submission two weeks before the end.
", "votes": 0, "replies": [] }, { "id": 676255, "author_name": "Hieu Phung", "author_url": "", "post_date": "2019-11-19T03:45:58.993000", "content": "Interesting ✨ . Thanks for sharing, and congratulations! 🎉
", "votes": 0, "replies": [] }, { "id": 676154, "author_name": "Salman", "author_url": "", "post_date": "2019-11-19T01:56:58.247000", "content": "Can you please share your code? I am really curious about lovasz loss working so well for this data.
", "votes": 0, "replies": [ { "id": 676161, "author_name": "Md Yasin Kabir", "author_url": "", "post_date": "2019-11-19T02:06:11.790000", "content": "I will share a writeup soon.
", "votes": 0, "replies": [] } ] }, { "id": 676152, "author_name": "liuze", "author_url": "", "post_date": "2019-11-19T01:54:33.513000", "content": "congratulation, where can we find @sgalib's lessons:)
", "votes": 0, "replies": [] }, { "id": 676084, "author_name": "hengck23", "author_url": "", "post_date": "2019-11-19T00:31:39.753000", "content": "thanks for the share. is it because of lovaz loss?
", "votes": 0, "replies": [ { "id": 676091, "author_name": "Md Yasin Kabir", "author_url": "", "post_date": "2019-11-19T00:37:35.667000", "content": "I think so.
", "votes": 1, "replies": [] }, { "id": 676096, "author_name": "Md Yasin Kabir", "author_url": "", "post_date": "2019-11-19T00:43:39.680000", "content": "I think the train and test dataset were similar. So the model gave the best CV scored best in private LB.
", "votes": 1, "replies": [] } ] }, { "id": 676345, "author_name": "", "author_url": "", "post_date": "2019-11-19T05:21:48.913000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "676075": "Congratulations to winners and golds. Also, congratulation to everyone who learns a lot from this contest. \n\nIn the last few days, I was getting funny results from different ensembles. Thus, I select low scored submissions (0.6732 and 0.6739) for finals. Guess what, the private LB came better for 0.6732 one. That's not the funny part. The most interesting part is in the following image. \n\n**My best score in private LB is 0.66758. But for the same submission public LB was 0.66654**. \nIt came from a single InceptionresnetV2 network 4-fold. \n\n\n\nFinally, I would like to give billions of thanks to @sgalib. Throughout this contest, I learn a lot from him. He encouraged me much to join in the kaggle competitions 2 years ago and taught me lots of ML concepts. I couldn't able to do this much without his cordial supports, lessons, and teaching. Personally, I am so grateful to him. \n", "677808": "Wow that's great @mykttu . Looking forward to your solution. ", "676086": "Nice. I couldn't fix my ensemble code but I still got bronze with only a single-fold efficientnetb2", "676080": "Wow, that's impressive. Will you share the write up about this single network? I look forward to see it :)", "678344": "Congratulations !........Great work.", "676285": "Sad story shared: I used to have my best submission two weeks before the end.", "676255": "Interesting ✨ . Thanks for sharing, and congratulations! 🎉 ", "676154": "Can you please share your code? I am really curious about lovasz loss working so well for this data.", "676152": "congratulation, where can we find @sgalib's lessons:)", "676084": "thanks for the share. is it because of lovaz loss?", "676345": "" } }