Shakeup was huge than expected but this was a great learning experience for me. I learnt a lot from @hengck23 code and discussions. I started with your starter code and improved on it, and the best public LB we got was 0.9898 and the best private LB was 0.9382. Thank you again for your contributions to this competition. \nAlso @haqishen, thank you so much for your sharings and discussions. For me, you dominated this competition, not just on LB but also in discussions(and kernels). Being in the top5 on LB and still contributing in discussions tells a lot about your confidence. Man, you are here to stay.\nI look forward to learn from you guys in many more competitions to come and Thank you once again for this one :)
Shakeup was huge than expected but this was a great learning experience for me. I learnt a lot from @hengck23 code and discussions. I started with your starter code and improved on it, and the best public LB we got was 0.9898 and the best private LB was 0.9382. Thank you again for your contributions to this competition. \nAlso @haqishen, thank you so much for your sharings and discussions. For me, you dominated this competition, not just on LB but also in discussions(and kernels). Being in the top5 on LB and still contributing in discussions tells a lot about your confidence. Man, you are here to stay.\nI look forward to learn from you guys in many more competitions to come and Thank you once again for this one :)
One thing that really surprises me about the frog brother of his deep analysis of the problem and top-level investigation. I have learned many new things from his works in just one competition and really looking forward to learning more cools stuff in upcoming competitions. Thank you @hengck23 sir for your great contribution, really glad to know about you. :)
\n\nlast but not least, you brother Bibek, you are just amazing. After when you shared about the randomly choosing mix-up and cutmix, people started to get high score within one week. like Qishen Ha told, you have saved an average of a week or two of time for many people ;) by your wonderful posts :)
", "rawMarkdown": "One thing that really surprises me about the frog brother of his deep analysis of the problem and top-level investigation. I have learned many new things from his works in just one competition and really looking forward to learning more cools stuff in upcoming competitions. Thank you @hengck23 sir for your great contribution, really glad to know about you. :) \n\nlast but not least, you brother Bibek, you are just amazing. After when you shared about the [randomly choosing mix-up and cutmix](https://www.kaggle.com/c/bengaliai-cv19/discussion/123198#718926), people started to get high score within one week. like Qishen Ha told, [**you have saved an average of a week or two of time for many people ;) by your wonderful posts**](https://www.kaggle.com/c/bengaliai-cv19/discussion/123198#719086) :)", "votes": null }, { "id": "775988", "postDate": "03/17/2020 04:11:41", "content": "Thank you for the kind words ;)
", "rawMarkdown": "Thank you for the kind words ;)", "votes": null }, { "id": "777493", "postDate": "03/17/2020 18:04:35", "content": "Bad that you missed the medal....Looks like this competition has the biggest shake up . Focus on Deep Fake challenge and win medal in that...Your ranking shows top 9 percent and why didnt you get a medal @bibek777
", "rawMarkdown": "Bad that you missed the medal....Looks like this competition has the biggest shake up . Focus on Deep Fake challenge and win medal in that...Your ranking shows top 9 percent and why didnt you get a medal @bibek777", "votes": null }, { "id": "778139", "postDate": "03/18/2020 07:19:42", "content": "Thanks to them.\nAnd thanks to you for sharing various amazing augmentations techniques. That led to instant leaderboard boosts.
", "rawMarkdown": "Thanks to them.\nAnd thanks to you for sharing various amazing augmentations techniques. That led to instant leaderboard boosts.", "votes": null }, { "id": "778167", "postDate": "03/18/2020 07:50:11", "content": "Thanks for the post.\nEveryone has done a fantastic job!
", "rawMarkdown": "Thanks for the post.\nEveryone has done a fantastic job!", "votes": null }, { "id": "778556", "postDate": "03/18/2020 14:58:52", "content": "Thank you @bibek777 ,@hengck23, @haqishen. you guys shared more than just information.
", "rawMarkdown": "Thank you @bibek777 ,@hengck23, @haqishen. you guys shared more than just information.", "votes": null }, { "id": "778597", "postDate": "03/18/2020 15:25:28", "content": "Seems that you didn't make a single submission till the end... 😂
", "rawMarkdown": "Seems that you didn't make a single submission till the end... 😂", "votes": null }, { "id": "778780", "postDate": "03/18/2020 18:14:37", "content": "@haqishen yes, I just kept testing various things you and other people shared on the forums on a subset of dataset (for quick computations). Honestly, it was hard enough for me to read so many papers and implement them by my own. I am really a CV newbie right now.\nAlso, I used colab for the most part and by the end I couldn't cross 0.968 so I never submitted xD. Although now I have learned a lot of things and am really excited for the next CV competition. Hopefully, I will have some compute resources by then.\nCongratulations on your medal though, I can't express how much have I learned from your posts and the final solution.
", "rawMarkdown": "haqishen yes, I just kept testing various things you and other people shared on the forums on a subset of dataset (for quick computations). Honestly, it was hard enough for me to read so many papers and implement them by my own. I am really a CV newbie right now.\nAlso, I used colab for the most part and by the end I couldn't cross 0.968 so I never submitted xD. Although now I have learned a lot of things and am really excited for the next CV competition. Hopefully, I will have some compute resources by then.\nCongratulations on your medal though, I can't express how much have I learned from your posts and the final solution.", "votes": null }, { "id": "778824", "postDate": "03/18/2020 19:14:01", "content": "Thank you @hengck23 @haqishen @MachineLP @bibek777 @iafoss for all the discussions and useful snippets of code =)
", "rawMarkdown": "Thank you @hengck23 @haqishen @MachineLP @bibek777 @iafoss for all the discussions and useful snippets of code =)", "votes": null } ], "comments": [ { "id": 775948, "author_name": "ipythonx", "author_url": "", "post_date": "03/17/2020 03:24:23", "content": "One thing that really surprises me about the frog brother of his deep analysis of the problem and top-level investigation. I have learned many new things from his works in just one competition and really looking forward to learning more cools stuff in upcoming competitions. Thank you @hengck23 sir for your great contribution, really glad to know about you. :)
\n\nlast but not least, you brother Bibek, you are just amazing. After when you shared about the randomly choosing mix-up and cutmix, people started to get high score within one week. like Qishen Ha told, you have saved an average of a week or two of time for many people ;) by your wonderful posts :)
", "votes": null, "replies": [] }, { "id": 775988, "author_name": "haqishen", "author_url": "", "post_date": "03/17/2020 04:11:41", "content": "Thank you for the kind words ;)
", "votes": null, "replies": [] }, { "id": 777493, "author_name": "usharengaraju", "author_url": "", "post_date": "03/17/2020 18:04:35", "content": "Bad that you missed the medal....Looks like this competition has the biggest shake up . Focus on Deep Fake challenge and win medal in that...Your ranking shows top 9 percent and why didnt you get a medal @bibek777
", "votes": null, "replies": [] }, { "id": 778139, "author_name": "timetraveller98", "author_url": "", "post_date": "03/18/2020 07:19:42", "content": "Thanks to them.\nAnd thanks to you for sharing various amazing augmentations techniques. That led to instant leaderboard boosts.
", "votes": null, "replies": [ { "id": 778597, "author_name": "haqishen", "author_url": "", "post_date": "03/18/2020 15:25:28", "content": "Seems that you didn't make a single submission till the end... 😂
", "votes": null, "replies": [] }, { "id": 778780, "author_name": "timetraveller98", "author_url": "", "post_date": "03/18/2020 18:14:37", "content": "@haqishen yes, I just kept testing various things you and other people shared on the forums on a subset of dataset (for quick computations). Honestly, it was hard enough for me to read so many papers and implement them by my own. I am really a CV newbie right now.\nAlso, I used colab for the most part and by the end I couldn't cross 0.968 so I never submitted xD. Although now I have learned a lot of things and am really excited for the next CV competition. Hopefully, I will have some compute resources by then.\nCongratulations on your medal though, I can't express how much have I learned from your posts and the final solution.
", "votes": null, "replies": [] } ] }, { "id": 778167, "author_name": "hengck23", "author_url": "", "post_date": "03/18/2020 07:50:11", "content": "Thanks for the post.\nEveryone has done a fantastic job!
", "votes": null, "replies": [] }, { "id": 778556, "author_name": "soonhwankwon", "author_url": "", "post_date": "03/18/2020 14:58:52", "content": "Thank you @bibek777 ,@hengck23, @haqishen. you guys shared more than just information.
", "votes": null, "replies": [] }, { "id": 778824, "author_name": "drhabib", "author_url": "", "post_date": "03/18/2020 19:14:01", "content": "Thank you @hengck23 @haqishen @MachineLP @bibek777 @iafoss for all the discussions and useful snippets of code =)
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "775935": "Shakeup was huge than expected but this was a great learning experience for me. I learnt a lot from @hengck23 code and discussions. I started with your starter code and improved on it, and the best public LB we got was `0.9898` and the best private LB was `0.9382`. Thank you again for your contributions to this competition. \nAlso @haqishen, thank you so much for your sharings and discussions. For me, you dominated this competition, not just on LB but also in discussions(and kernels). Being in the top5 on LB and still contributing in discussions tells a lot about your confidence. Man, you are here to stay.\nI look forward to learn from you guys in many more competitions to come and Thank you once again for this one :)", "775948": "One thing that really surprises me about the frog brother of his deep analysis of the problem and top-level investigation. I have learned many new things from his works in just one competition and really looking forward to learning more cools stuff in upcoming competitions. Thank you @hengck23 sir for your great contribution, really glad to know about you. :) \n\nlast but not least, you brother Bibek, you are just amazing. After when you shared about the [randomly choosing mix-up and cutmix](https://www.kaggle.com/c/bengaliai-cv19/discussion/123198#718926), people started to get high score within one week. like Qishen Ha told, [**you have saved an average of a week or two of time for many people ;) by your wonderful posts**](https://www.kaggle.com/c/bengaliai-cv19/discussion/123198#719086) :)", "775988": "Thank you for the kind words ;)", "777493": "Bad that you missed the medal....Looks like this competition has the biggest shake up . Focus on Deep Fake challenge and win medal in that...Your ranking shows top 9 percent and why didnt you get a medal @bibek777", "778139": "Thanks to them.\nAnd thanks to you for sharing various amazing augmentations techniques. That led to instant leaderboard boosts.", "778167": "Thanks for the post.\nEveryone has done a fantastic job!", "778556": "Thank you @bibek777 ,@hengck23, @haqishen. you guys shared more than just information.", "778597": "Seems that you didn't make a single submission till the end... 😂", "778780": "haqishen yes, I just kept testing various things you and other people shared on the forums on a subset of dataset (for quick computations). Honestly, it was hard enough for me to read so many papers and implement them by my own. I am really a CV newbie right now.\nAlso, I used colab for the most part and by the end I couldn't cross 0.968 so I never submitted xD. Although now I have learned a lot of things and am really excited for the next CV competition. Hopefully, I will have some compute resources by then.\nCongratulations on your medal though, I can't express how much have I learned from your posts and the final solution.", "778824": "Thank you @hengck23 @haqishen @MachineLP @bibek777 @iafoss for all the discussions and useful snippets of code =)" }, "source": "meta" }