{ "id": 242509, "title": "Ready for final sprint?", "url": "/competitions/birdclef-2021/discussion/242509", "author_name": "", "post_date": "2021-05-29T10:48:36.062846400Z", "votes": 31, "comment_count": 42, "views": 0, "content": "

We had our drama with the submission outage. I hope it did not disturb you too much.

\n

The LB is heating up. As I predicted there will be several teams above 0.80 on public LB. Final will be intense.

\n

Don't panic if you see others climbing the LB, this is to be expected. Check your sub twice before submitting. And do your best!

\n

I will certainly try to do my best till the end. Leading the way till last moment is not the same as winning. Only final ranking will stay.

", "messages": [ { "id": "1327493", "postDate": "05/29/2021 10:48:36", "content": "

We had our drama with the submission outage. I hope it did not disturb you too much.

\n

The LB is heating up. As I predicted there will be several teams above 0.80 on public LB. Final will be intense.

\n

Don't panic if you see others climbing the LB, this is to be expected. Check your sub twice before submitting. And do your best!

\n

I will certainly try to do my best till the end. Leading the way till last moment is not the same as winning. Only final ranking will stay.

", "rawMarkdown": "We had our drama with the submission outage. I hope it did not disturb you too much.\n\nThe LB is heating up. As I predicted there will be several teams above 0.80 on public LB. Final will be intense.\n\nDon't panic if you see others climbing the LB, this is to be expected. Check your sub twice before submitting. And do your best! \n\nI will certainly try to do my best till the end. Leading the way till last moment is not the same as winning. Only final ranking will stay.", "votes": null }, { "id": "1327544", "postDate": "05/29/2021 12:05:43", "content": "

LOL, some are really too stressed (downvoted).

", "rawMarkdown": "LOL, some are really too stressed (downvoted).", "votes": null }, { "id": "1327545", "postDate": "05/29/2021 12:07:04", "content": "

Good luck CPMP!! Indeed will be very intense days…

\n

ps: IfI may add one suggestion to the above: Select wisely your final subs ;) This is what matters at the end

", "rawMarkdown": "Good luck CPMP!! Indeed will be very intense days... \n\nps: IfI may add one suggestion to the above: Select wisely your final subs ;) This is what matters at the end", "votes": null }, { "id": "1327553", "postDate": "05/29/2021 12:12:39", "content": "

Thanks. Good luck to you as well.

\n

You are right, final selection is key.

", "rawMarkdown": "Thanks. Good luck to you as well.\n\nYou are right, final selection is key.", "votes": null }, { "id": "1327571", "postDate": "05/29/2021 12:32:43", "content": "

I don't really get the upvote ecosystem on Kaggle.

\n

On a related note, I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not…

", "rawMarkdown": "I don't really get the upvote ecosystem on Kaggle.\n\nOn a related note, I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not...", "votes": null }, { "id": "1327621", "postDate": "05/29/2021 13:13:02", "content": "
\n

I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not…

\n
\n

Same here. Even though I have been discussion GM way before competition GM and enjoyed the (discussion) GM title.

", "rawMarkdown": "> I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not… \n\nSame here. Even though I have been discussion GM way before competition GM and enjoyed the (discussion) GM title.", "votes": null }, { "id": "1327631", "postDate": "05/29/2021 13:29:02", "content": "

I also look at the total ranking, maybe I’m way too biased in the matter as the highest ranked Kaggle Expert 😉

", "rawMarkdown": "I also look at the total ranking, maybe I’m way too biased in the matter as the highest ranked Kaggle Expert 😉", "votes": null }, { "id": "1327639", "postDate": "05/29/2021 13:37:05", "content": "

Great work! I quote myself from another competition

\n
\n

\"It ain't over till mr @cpmpml sings\".

\n
", "rawMarkdown": "Great work! I quote myself from another competition\n> \"It ain't over till mr @cpmpml sings\".", "votes": null }, { "id": "1327657", "postDate": "05/29/2021 14:04:24", "content": "

I missed it! Where did you wrote this first?

", "rawMarkdown": "I missed it! Where did you wrote this first?", "votes": null }, { "id": "1327665", "postDate": "05/29/2021 14:10:46", "content": "

Its was not recently, this old quote started long back in the days, ;) in the OpenVaccine: COVID-19 mRNA Vaccine Degradation Prediction 👍

", "rawMarkdown": "Its was not recently, this old quote started long back in the days, ;) in the OpenVaccine: COVID-19 mRNA Vaccine Degradation Prediction 👍", "votes": null }, { "id": "1327756", "postDate": "05/29/2021 15:47:39", "content": "

Good luck everyone! I hope to give my best.

\n

Everyone is pretty close, there is decent chance of small shake up as well.

", "rawMarkdown": "Good luck everyone! I hope to give my best. \n\nEveryone is pretty close, there is decent chance of small shake up as well.", "votes": null }, { "id": "1327864", "postDate": "05/29/2021 17:26:11", "content": "

You don't need to do your best, you're high enough ;) relax and enjoy life!

", "rawMarkdown": "You don't need to do your best, you're high enough ;) relax and enjoy life!", "votes": null }, { "id": "1327892", "postDate": "05/29/2021 17:58:31", "content": "

Nice try :D

", "rawMarkdown": "Nice try :D", "votes": null }, { "id": "1327916", "postDate": "05/29/2021 18:26:53", "content": "

What about you both take a break?

", "rawMarkdown": "What about you both take a break?", "votes": null }, { "id": "1327923", "postDate": "05/29/2021 18:31:53", "content": "

I agree. Make way for the young!

", "rawMarkdown": "I agree. Make way for the young!", "votes": null }, { "id": "1327966", "postDate": "05/29/2021 19:26:27", "content": "

I came here 10-11 days prior and it was so calm here , 4-5 days in the competition it lit up like winter fire , the score differences are not that huge , its all about the single idea/model kicking , its going to be really tough sleepless nights till Tuesday

\n

Best of luck to you @cpmpml

", "rawMarkdown": "I came here 10-11 days prior and it was so calm here , 4-5 days in the competition it lit up like winter fire , the score differences are not that huge , its all about the single idea/model kicking , its going to be really tough sleepless nights till Tuesday\n\nBest of luck to you @cpmpml", "votes": null }, { "id": "1328006", "postDate": "05/29/2021 20:50:16", "content": "
\n

I agree. Make way for the young!

\n
\n

@christofhenkel may want a word on that

", "rawMarkdown": ">I agree. Make way for the young!\n\n@christofhenkel may want a word on that", "votes": null }, { "id": "1328057", "postDate": "05/29/2021 22:49:48", "content": "

Thanks! I knew it would end up tight. An extra day makes it even more likely.

", "rawMarkdown": "Thanks! I knew it would end up tight. An extra day makes it even more likely.", "votes": null }, { "id": "1328180", "postDate": "05/30/2021 02:36:35", "content": "
\n

Everyone is pretty close, there is decent chance of small shake up as well.

\n
\n

Do you guys expect a small shake-up? I think the shake-up will be +-0.07 at where I am but it may be different for the gold zone.

", "rawMarkdown": "> Everyone is pretty close, there is decent chance of small shake up as well.\n\nDo you guys expect a small shake-up? I think the shake-up will be +-0.07 at where I am but it may be different for the gold zone.", "votes": null }, { "id": "1328254", "postDate": "05/30/2021 04:55:38", "content": "

@cpmpml i hope you reclaim your month long tittle back at private board listing . what is your opinion about the public kernels scoring 0.6 + . are they overfit which i think they are..

", "rawMarkdown": "cpmpml i hope you reclaim your month long tittle back at private board listing . what is your opinion about the public kernels scoring 0.6 + . are they overfit which i think they are..", "votes": null }, { "id": "1328275", "postDate": "05/30/2021 05:41:24", "content": "

@jaideepvalani why do you think they are overfit?

", "rawMarkdown": "jaideepvalani why do you think they are overfit?", "votes": null }, { "id": "1328382", "postDate": "05/30/2021 07:59:43", "content": "

Why do you think that shake-up will be +-0.07? Till now, my CV-LB correlation is almost perfect!! I feel the shakeup may be in range +-0.02, or, in worst case +-0.03/0.04 !!

", "rawMarkdown": "Why do you think that shake-up will be +-0.07? Till now, my CV-LB correlation is almost perfect!! I feel the shakeup may be in range +-0.02, or, in worst case +-0.03/0.04 !!", "votes": null }, { "id": "1328406", "postDate": "05/30/2021 08:18:56", "content": "

For me CV/LB = 0.8/0.78 and correlates.
\nThe score will be different if the private data has a different percentage of nocall labels, but I don't expect a shake-up

", "rawMarkdown": "For me CV/LB = 0.8/0.78 and correlates. \nThe score will be different if the private data has a different percentage of nocall labels, but I don't expect a shake-up", "votes": null }, { "id": "1328508", "postDate": "05/30/2021 10:06:29", "content": "

Guys by CV you mean the train CV (on short audios) or with Train SC ?
\nI agree that the nocall/bird call distribution would play a role

", "rawMarkdown": "Guys by CV you mean the train CV (on short audios) or with Train SC ? \nI agree that the nocall/bird call distribution would play a role", "votes": null }, { "id": "1328534", "postDate": "05/30/2021 10:35:56", "content": "

I am calculating CV on train soundscapes! For me CV/LB = 0.803/0.77!! I keep a check on both, CV on call samples as well as CV on nocall samples!! You can get same CV in 2 submissions where one of them have more call F1 and other may have more nocall F1!! I agree that the scores will be different depending on distribution, but there can be small shakeup in range +- 0.01/0.02 (relative to final scores) depending on whether your submission does better on call samples or nocall samples !!

", "rawMarkdown": "I am calculating CV on train soundscapes! For me CV/LB = 0.803/0.77!! I keep a check on both, CV on call samples as well as CV on nocall samples!! You can get same CV in 2 submissions where one of them have more call F1 and other may have more nocall F1!! I agree that the scores will be different depending on distribution, but there can be small shakeup in range +- 0.01/0.02 (relative to final scores) depending on whether your submission does better on call samples or nocall samples !!", "votes": null }, { "id": "1328538", "postDate": "05/30/2021 10:38:40", "content": "

@aerdem4 I would also guess +/- 0.02 based on manual inspection of train soundscapes, unless for some reason couple of species not predicted well by model dominate hidden test set.

", "rawMarkdown": "aerdem4 I would also guess +/- 0.02 based on manual inspection of train soundscapes, unless for some reason couple of species not predicted well by model dominate hidden test set.", "votes": null }, { "id": "1328582", "postDate": "05/30/2021 11:28:37", "content": "

may be the most simplistic method use .. i mean it can be any way around . Public kernel have history of outclassing several other models .

", "rawMarkdown": "may be the most simplistic method use .. i mean it can be any way around . Public kernel have history of outclassing several other models .", "votes": null }, { "id": "1328596", "postDate": "05/30/2021 11:50:10", "content": "

@jaideepvalani I have not looked at public notebooks in this competition, sorry.

", "rawMarkdown": "jaideepvalani I have not looked at public notebooks in this competition, sorry.", "votes": null }, { "id": "1328646", "postDate": "05/30/2021 12:20:00", "content": "
\n

i hope you reclaim your month long tittle back at private board listing .

\n
\n

@jaideepvalani Thanks!

", "rawMarkdown": "> i hope you reclaim your month long tittle back at private board listing .\n\n@jaideepvalani Thanks!", "votes": null }, { "id": "1329326", "postDate": "05/31/2021 04:05:42", "content": "

What CV are you doing for validation? I have been using a leave-one-soundscape-out CV but this can take a long time to run.

", "rawMarkdown": "What CV are you doing for validation? I have been using a leave-one-soundscape-out CV but this can take a long time to run.", "votes": null }, { "id": "1329644", "postDate": "05/31/2021 09:15:21", "content": "

After several weeks of procrastination last week we finally decided with @pestipeti to let this competition go :D

\n

Best of luck everyone! I am still interested reading the solution write-ups

", "rawMarkdown": "After several weeks of procrastination last week we finally decided with @pestipeti to let this competition go :D\n\nBest of luck everyone! I am still interested reading the solution write-ups", "votes": null }, { "id": "1329855", "postDate": "05/31/2021 12:26:44", "content": "

@tanulsingh077 Winter fires can be wild - especially in the summer ;) :P

", "rawMarkdown": "tanulsingh077 Winter fires can be wild - especially in the summer ;) :P", "votes": null }, { "id": "1329889", "postDate": "05/31/2021 12:56:12", "content": "

I was waiting for your team coming out of the blue near or at the top…

", "rawMarkdown": "I was waiting for your team coming out of the blue near or at the top...", "votes": null }, { "id": "1329900", "postDate": "05/31/2021 13:03:43", "content": "

I can promise we won't do that :D

", "rawMarkdown": "I can promise we won't do that :D", "votes": null }, { "id": "1329901", "postDate": "05/31/2021 13:04:24", "content": "

I had same impression - that you guys are holding a surprise for the last days :)
\nFor sure your contribution/approach will be missing from this one

", "rawMarkdown": "I had same impression - that you guys are holding a surprise for the last days :) \nFor sure your contribution/approach will be missing from this one", "votes": null }, { "id": "1331556", "postDate": "06/01/2021 14:36:34", "content": "

Procrastinating to let it go. Funny :D

", "rawMarkdown": "Procrastinating to let it go. Funny :D", "votes": null }, { "id": "1331803", "postDate": "06/01/2021 18:21:48", "content": "

It's a tight race

", "rawMarkdown": "It's a tight race", "votes": null }, { "id": "1332039", "postDate": "06/01/2021 21:30:16", "content": "

We don't know that yet. We will will know once the private LB is revealed.

", "rawMarkdown": "We don't know that yet. We will will know once the private LB is revealed.", "votes": null }, { "id": "1332042", "postDate": "06/01/2021 21:33:27", "content": "

classes are less that is another point of shake up..

", "rawMarkdown": "classes are less that is another point of shake up..", "votes": null }, { "id": "1332057", "postDate": "06/01/2021 21:47:59", "content": "

no idea how you could work on that competition for so long, very exhausting °_°

", "rawMarkdown": "no idea how you could work on that competition for so long, very exhausting °_°", "votes": null }, { "id": "1332123", "postDate": "06/01/2021 23:39:55", "content": "

Well, your team and few other competitors made is spicy till the end for me ;)

", "rawMarkdown": "Well, your team and few other competitors made is spicy till the end for me ;)", "votes": null }, { "id": "1332595", "postDate": "06/02/2021 07:51:33", "content": "

I was expecting this shake-up but since you all said your CV and LB are correlated, I thought I was wrong.

", "rawMarkdown": "I was expecting this shake-up but since you all said your CV and LB are correlated, I thought I was wrong.", "votes": null }, { "id": "1332646", "postDate": "06/02/2021 08:22:56", "content": "

Yes it’s was the same in the previous birdcall competition, many movements, so it was expected.
\nPersonally, I went from #102 in the previous to #54 in this one, maybe I have learnt something since the last competition 9 months ago, got more knowledge in AI/Data Science/python and now some experience within audio, or just luck, one cannot be sure 😉 but I vote and believe on the increasing of knowledge.

", "rawMarkdown": "Yes it’s was the same in the previous birdcall competition, many movements, so it was expected.\nPersonally, I went from #102 in the previous to #54 in this one, maybe I have learnt something since the last competition 9 months ago, got more knowledge in AI/Data Science/python and now some experience within audio, or just luck, one cannot be sure 😉 but I vote and believe on the increasing of knowledge.", "votes": null } ], "comments": [ { "id": 1327544, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 12:05:43", "content": "

LOL, some are really too stressed (downvoted).

", "votes": null, "replies": [ { "id": 1327571, "author_name": "leecming", "author_url": "", "post_date": "05/29/2021 12:32:43", "content": "

I don't really get the upvote ecosystem on Kaggle.

\n

On a related note, I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not…

", "votes": null, "replies": [] }, { "id": 1327621, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 13:13:02", "content": "
\n

I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not…

\n
\n

Same here. Even though I have been discussion GM way before competition GM and enjoyed the (discussion) GM title.

", "votes": null, "replies": [] }, { "id": 1327631, "author_name": "kirderf", "author_url": "", "post_date": "05/29/2021 13:29:02", "content": "

I also look at the total ranking, maybe I’m way too biased in the matter as the highest ranked Kaggle Expert 😉

", "votes": null, "replies": [] }, { "id": 1328254, "author_name": "jaideepvalani", "author_url": "", "post_date": "05/30/2021 04:55:38", "content": "

@cpmpml i hope you reclaim your month long tittle back at private board listing . what is your opinion about the public kernels scoring 0.6 + . are they overfit which i think they are..

", "votes": null, "replies": [] }, { "id": 1328275, "author_name": "aerdem4", "author_url": "", "post_date": "05/30/2021 05:41:24", "content": "

@jaideepvalani why do you think they are overfit?

", "votes": null, "replies": [] }, { "id": 1328582, "author_name": "jaideepvalani", "author_url": "", "post_date": "05/30/2021 11:28:37", "content": "

may be the most simplistic method use .. i mean it can be any way around . Public kernel have history of outclassing several other models .

", "votes": null, "replies": [] }, { "id": 1328596, "author_name": "cpmpml", "author_url": "", "post_date": "05/30/2021 11:50:10", "content": "

@jaideepvalani I have not looked at public notebooks in this competition, sorry.

", "votes": null, "replies": [] }, { "id": 1328646, "author_name": "cpmpml", "author_url": "", "post_date": "05/30/2021 12:20:00", "content": "
\n

i hope you reclaim your month long tittle back at private board listing .

\n
\n

@jaideepvalani Thanks!

", "votes": null, "replies": [] } ] }, { "id": 1327545, "author_name": "imeintanis", "author_url": "", "post_date": "05/29/2021 12:07:04", "content": "

Good luck CPMP!! Indeed will be very intense days…

\n

ps: IfI may add one suggestion to the above: Select wisely your final subs ;) This is what matters at the end

", "votes": null, "replies": [ { "id": 1327553, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 12:12:39", "content": "

Thanks. Good luck to you as well.

\n

You are right, final selection is key.

", "votes": null, "replies": [] } ] }, { "id": 1327639, "author_name": "kirderf", "author_url": "", "post_date": "05/29/2021 13:37:05", "content": "

Great work! I quote myself from another competition

\n
\n

\"It ain't over till mr @cpmpml sings\".

\n
", "votes": null, "replies": [ { "id": 1327657, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 14:04:24", "content": "

I missed it! Where did you wrote this first?

", "votes": null, "replies": [] }, { "id": 1327665, "author_name": "kirderf", "author_url": "", "post_date": "05/29/2021 14:10:46", "content": "

Its was not recently, this old quote started long back in the days, ;) in the OpenVaccine: COVID-19 mRNA Vaccine Degradation Prediction 👍

", "votes": null, "replies": [] } ] }, { "id": 1327756, "author_name": "tezdhar", "author_url": "", "post_date": "05/29/2021 15:47:39", "content": "

Good luck everyone! I hope to give my best.

\n

Everyone is pretty close, there is decent chance of small shake up as well.

", "votes": null, "replies": [ { "id": 1327864, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 17:26:11", "content": "

You don't need to do your best, you're high enough ;) relax and enjoy life!

", "votes": null, "replies": [] }, { "id": 1327892, "author_name": "tezdhar", "author_url": "", "post_date": "05/29/2021 17:58:31", "content": "

Nice try :D

", "votes": null, "replies": [] }, { "id": 1327916, "author_name": "philippsinger", "author_url": "", "post_date": "05/29/2021 18:26:53", "content": "

What about you both take a break?

", "votes": null, "replies": [] }, { "id": 1327923, "author_name": "vlomme", "author_url": "", "post_date": "05/29/2021 18:31:53", "content": "

I agree. Make way for the young!

", "votes": null, "replies": [] }, { "id": 1328006, "author_name": "ilu000", "author_url": "", "post_date": "05/29/2021 20:50:16", "content": "
\n

I agree. Make way for the young!

\n
\n

@christofhenkel may want a word on that

", "votes": null, "replies": [] }, { "id": 1328180, "author_name": "aerdem4", "author_url": "", "post_date": "05/30/2021 02:36:35", "content": "
\n

Everyone is pretty close, there is decent chance of small shake up as well.

\n
\n

Do you guys expect a small shake-up? I think the shake-up will be +-0.07 at where I am but it may be different for the gold zone.

", "votes": null, "replies": [] }, { "id": 1328382, "author_name": "ks2019", "author_url": "", "post_date": "05/30/2021 07:59:43", "content": "

Why do you think that shake-up will be +-0.07? Till now, my CV-LB correlation is almost perfect!! I feel the shakeup may be in range +-0.02, or, in worst case +-0.03/0.04 !!

", "votes": null, "replies": [] }, { "id": 1328406, "author_name": "vlomme", "author_url": "", "post_date": "05/30/2021 08:18:56", "content": "

For me CV/LB = 0.8/0.78 and correlates.
\nThe score will be different if the private data has a different percentage of nocall labels, but I don't expect a shake-up

", "votes": null, "replies": [] }, { "id": 1328508, "author_name": "imeintanis", "author_url": "", "post_date": "05/30/2021 10:06:29", "content": "

Guys by CV you mean the train CV (on short audios) or with Train SC ?
\nI agree that the nocall/bird call distribution would play a role

", "votes": null, "replies": [] }, { "id": 1328534, "author_name": "ks2019", "author_url": "", "post_date": "05/30/2021 10:35:56", "content": "

I am calculating CV on train soundscapes! For me CV/LB = 0.803/0.77!! I keep a check on both, CV on call samples as well as CV on nocall samples!! You can get same CV in 2 submissions where one of them have more call F1 and other may have more nocall F1!! I agree that the scores will be different depending on distribution, but there can be small shakeup in range +- 0.01/0.02 (relative to final scores) depending on whether your submission does better on call samples or nocall samples !!

", "votes": null, "replies": [] }, { "id": 1328538, "author_name": "tezdhar", "author_url": "", "post_date": "05/30/2021 10:38:40", "content": "

@aerdem4 I would also guess +/- 0.02 based on manual inspection of train soundscapes, unless for some reason couple of species not predicted well by model dominate hidden test set.

", "votes": null, "replies": [] }, { "id": 1329326, "author_name": "aarons0", "author_url": "", "post_date": "05/31/2021 04:05:42", "content": "

What CV are you doing for validation? I have been using a leave-one-soundscape-out CV but this can take a long time to run.

", "votes": null, "replies": [] }, { "id": 1332042, "author_name": "jaideepvalani", "author_url": "", "post_date": "06/01/2021 21:33:27", "content": "

classes are less that is another point of shake up..

", "votes": null, "replies": [] }, { "id": 1332595, "author_name": "aerdem4", "author_url": "", "post_date": "06/02/2021 07:51:33", "content": "

I was expecting this shake-up but since you all said your CV and LB are correlated, I thought I was wrong.

", "votes": null, "replies": [] }, { "id": 1332646, "author_name": "kirderf", "author_url": "", "post_date": "06/02/2021 08:22:56", "content": "

Yes it’s was the same in the previous birdcall competition, many movements, so it was expected.
\nPersonally, I went from #102 in the previous to #54 in this one, maybe I have learnt something since the last competition 9 months ago, got more knowledge in AI/Data Science/python and now some experience within audio, or just luck, one cannot be sure 😉 but I vote and believe on the increasing of knowledge.

", "votes": null, "replies": [] } ] }, { "id": 1327966, "author_name": "tanulsingh077", "author_url": "", "post_date": "05/29/2021 19:26:27", "content": "

I came here 10-11 days prior and it was so calm here , 4-5 days in the competition it lit up like winter fire , the score differences are not that huge , its all about the single idea/model kicking , its going to be really tough sleepless nights till Tuesday

\n

Best of luck to you @cpmpml

", "votes": null, "replies": [ { "id": 1328057, "author_name": "cpmpml", "author_url": "", "post_date": "05/29/2021 22:49:48", "content": "

Thanks! I knew it would end up tight. An extra day makes it even more likely.

", "votes": null, "replies": [] }, { "id": 1329855, "author_name": "killimi", "author_url": "", "post_date": "05/31/2021 12:26:44", "content": "

@tanulsingh077 Winter fires can be wild - especially in the summer ;) :P

", "votes": null, "replies": [] } ] }, { "id": 1329644, "author_name": "gaborfodor", "author_url": "", "post_date": "05/31/2021 09:15:21", "content": "

After several weeks of procrastination last week we finally decided with @pestipeti to let this competition go :D

\n

Best of luck everyone! I am still interested reading the solution write-ups

", "votes": null, "replies": [ { "id": 1329889, "author_name": "cpmpml", "author_url": "", "post_date": "05/31/2021 12:56:12", "content": "

I was waiting for your team coming out of the blue near or at the top…

", "votes": null, "replies": [] }, { "id": 1329900, "author_name": "gaborfodor", "author_url": "", "post_date": "05/31/2021 13:03:43", "content": "

I can promise we won't do that :D

", "votes": null, "replies": [] }, { "id": 1329901, "author_name": "imeintanis", "author_url": "", "post_date": "05/31/2021 13:04:24", "content": "

I had same impression - that you guys are holding a surprise for the last days :)
\nFor sure your contribution/approach will be missing from this one

", "votes": null, "replies": [] }, { "id": 1331556, "author_name": "nofreewill", "author_url": "", "post_date": "06/01/2021 14:36:34", "content": "

Procrastinating to let it go. Funny :D

", "votes": null, "replies": [] } ] }, { "id": 1331803, "author_name": "taoyi07", "author_url": "", "post_date": "06/01/2021 18:21:48", "content": "

It's a tight race

", "votes": null, "replies": [ { "id": 1332039, "author_name": "cpmpml", "author_url": "", "post_date": "06/01/2021 21:30:16", "content": "

We don't know that yet. We will will know once the private LB is revealed.

", "votes": null, "replies": [] } ] }, { "id": 1332057, "author_name": "philippsinger", "author_url": "", "post_date": "06/01/2021 21:47:59", "content": "

no idea how you could work on that competition for so long, very exhausting °_°

", "votes": null, "replies": [ { "id": 1332123, "author_name": "cpmpml", "author_url": "", "post_date": "06/01/2021 23:39:55", "content": "

Well, your team and few other competitors made is spicy till the end for me ;)

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1327493": "We had our drama with the submission outage. I hope it did not disturb you too much.\n\nThe LB is heating up. As I predicted there will be several teams above 0.80 on public LB. Final will be intense.\n\nDon't panic if you see others climbing the LB, this is to be expected. Check your sub twice before submitting. And do your best! \n\nI will certainly try to do my best till the end. Leading the way till last moment is not the same as winning. Only final ranking will stay.", "1327544": "LOL, some are really too stressed (downvoted).", "1327545": "Good luck CPMP!! Indeed will be very intense days... \n\nps: IfI may add one suggestion to the above: Select wisely your final subs ;) This is what matters at the end", "1327553": "Thanks. Good luck to you as well.\n\nYou are right, final selection is key.", "1327571": "I don't really get the upvote ecosystem on Kaggle.\n\nOn a related note, I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not...", "1327621": "> I'm pretty skeptical of the notebook/discussion tiers including Grandmaster. I see \"Kaggle Master\" on CVs quite frequently these days and I have to check whether it's a competition master or not… \n\nSame here. Even though I have been discussion GM way before competition GM and enjoyed the (discussion) GM title.", "1327631": "I also look at the total ranking, maybe I’m way too biased in the matter as the highest ranked Kaggle Expert 😉", "1327639": "Great work! I quote myself from another competition\n> \"It ain't over till mr @cpmpml sings\".", "1327657": "I missed it! Where did you wrote this first?", "1327665": "Its was not recently, this old quote started long back in the days, ;) in the OpenVaccine: COVID-19 mRNA Vaccine Degradation Prediction 👍", "1327756": "Good luck everyone! I hope to give my best. \n\nEveryone is pretty close, there is decent chance of small shake up as well.", "1327864": "You don't need to do your best, you're high enough ;) relax and enjoy life!", "1327892": "Nice try :D", "1327916": "What about you both take a break?", "1327923": "I agree. Make way for the young!", "1327966": "I came here 10-11 days prior and it was so calm here , 4-5 days in the competition it lit up like winter fire , the score differences are not that huge , its all about the single idea/model kicking , its going to be really tough sleepless nights till Tuesday\n\nBest of luck to you @cpmpml", "1328006": ">I agree. Make way for the young!\n\n@christofhenkel may want a word on that", "1328057": "Thanks! I knew it would end up tight. An extra day makes it even more likely.", "1328180": "> Everyone is pretty close, there is decent chance of small shake up as well.\n\nDo you guys expect a small shake-up? I think the shake-up will be +-0.07 at where I am but it may be different for the gold zone.", "1328254": "cpmpml i hope you reclaim your month long tittle back at private board listing . what is your opinion about the public kernels scoring 0.6 + . are they overfit which i think they are..", "1328275": "jaideepvalani why do you think they are overfit?", "1328382": "Why do you think that shake-up will be +-0.07? Till now, my CV-LB correlation is almost perfect!! I feel the shakeup may be in range +-0.02, or, in worst case +-0.03/0.04 !!", "1328406": "For me CV/LB = 0.8/0.78 and correlates. \nThe score will be different if the private data has a different percentage of nocall labels, but I don't expect a shake-up", "1328508": "Guys by CV you mean the train CV (on short audios) or with Train SC ? \nI agree that the nocall/bird call distribution would play a role", "1328534": "I am calculating CV on train soundscapes! For me CV/LB = 0.803/0.77!! I keep a check on both, CV on call samples as well as CV on nocall samples!! You can get same CV in 2 submissions where one of them have more call F1 and other may have more nocall F1!! I agree that the scores will be different depending on distribution, but there can be small shakeup in range +- 0.01/0.02 (relative to final scores) depending on whether your submission does better on call samples or nocall samples !!", "1328538": "aerdem4 I would also guess +/- 0.02 based on manual inspection of train soundscapes, unless for some reason couple of species not predicted well by model dominate hidden test set.", "1328582": "may be the most simplistic method use .. i mean it can be any way around . Public kernel have history of outclassing several other models .", "1328596": "jaideepvalani I have not looked at public notebooks in this competition, sorry.", "1328646": "> i hope you reclaim your month long tittle back at private board listing .\n\n@jaideepvalani Thanks!", "1329326": "What CV are you doing for validation? I have been using a leave-one-soundscape-out CV but this can take a long time to run.", "1329644": "After several weeks of procrastination last week we finally decided with @pestipeti to let this competition go :D\n\nBest of luck everyone! I am still interested reading the solution write-ups", "1329855": "tanulsingh077 Winter fires can be wild - especially in the summer ;) :P", "1329889": "I was waiting for your team coming out of the blue near or at the top...", "1329900": "I can promise we won't do that :D", "1329901": "I had same impression - that you guys are holding a surprise for the last days :) \nFor sure your contribution/approach will be missing from this one", "1331556": "Procrastinating to let it go. Funny :D", "1331803": "It's a tight race", "1332039": "We don't know that yet. We will will know once the private LB is revealed.", "1332042": "classes are less that is another point of shake up..", "1332057": "no idea how you could work on that competition for so long, very exhausting °_°", "1332123": "Well, your team and few other competitors made is spicy till the end for me ;)", "1332595": "I was expecting this shake-up but since you all said your CV and LB are correlated, I thought I was wrong.", "1332646": "Yes it’s was the same in the previous birdcall competition, many movements, so it was expected.\nPersonally, I went from #102 in the previous to #54 in this one, maybe I have learnt something since the last competition 9 months ago, got more knowledge in AI/Data Science/python and now some experience within audio, or just luck, one cannot be sure 😉 but I vote and believe on the increasing of knowledge." }, "source": "meta" }