Just want to make sure the ensemble won't exceed the time limit.
", "messages": [ { "id": 1789511, "postDate": "2022-05-14T01:04:01.757Z", "content": "Just to clarify the solution helps me reach #1 in public lb is not an ensemble of some models, the solution is pretty strange and contains some external resources under non-commercial license, not sure if the host will take it. Honestly if it is not almost the end of the competition, I will make the solution public available, but now I will publish this solution right after the competition ends. I will also make one of my final submissions contains ensemble of several models that were only trained on this competition data, which should rank around #25 currently.
", "rawMarkdown": "Just to clarify the solution helps me reach #1 in public lb is not an ensemble of some models, the solution is pretty strange and contains some external resources under non-commercial license, not sure if the host will take it. Honestly if it is not almost the end of the competition, I will make the solution public available, but now I will publish this solution right after the competition ends. I will also make one of my final submissions contains ensemble of several models that were only trained on this competition data, which should rank around #25 currently.", "votes": 8, "replies": [ { "id": 1789579, "postDate": "2022-05-14T03:05:02.483Z", "content": "I suggest you to read the rules regarding external data and this thread again.
\nHosts don't allow to use data with No Derivatives licenses. For example, this is rare species in Hawaii but is not included in the training data as it is granted No Derivatives licenses. We can't use data in the Macaulay Library either.
@leonshangguan
\nFYI: note that the mentioning about external data by competition host:
\nhttps://www.kaggle.com/competitions/birdclef-2022/discussion/315361#1737178
@leonshangguan I doubt that you can back your final solution with a second submission that does not contain unapproved external data.
\nIf your best submission contains 'illegal' data, I think you'll simply get removed from the LB, you won't get down to your second best submission.
@leonshangguan ok, spit it out man, what's the weird solution. Is it template matching ? Ivector?
", "rawMarkdown": "@leonshangguan ok, spit it out man, what's the weird solution. Is it template matching ? Ivector? " }, { "id": 1800453, "postDate": "2022-05-25T01:37:10.013Z", "content": "Working on writing a discussion bro, give me 1h. But you can see our inference notebooks here and here. Promise me don't say the \"F\" word after seeing the solution.
", "rawMarkdown": "Working on writing a discussion bro, give me 1h. But you can see our inference notebooks [here](https://www.kaggle.com/code/leonshangguan/private-7-8-final-of-submission) and [here](https://www.kaggle.com/code/leonshangguan/birdnet-inference). Promise me don't say the \"F\" word after seeing the solution.", "votes": 3 }, { "id": 1800484, "postDate": "2022-05-25T02:08:56.943Z", "content": "haha, not at all. Thanks for satisfying my curiosity.
\nIt teaches me to do a more proper literature study. Well, next time! Congratulations by the way, awesome result !
Looks like your ensemble is doing alright 😂
\nCongrats!
approximately 5,500 recordings
\nhttps://www.kaggle.com/competitions/birdclef-2022/data
", "rawMarkdown": "approximately 5,500 recordings\n\nhttps://www.kaggle.com/competitions/birdclef-2022/data", "votes": 2, "replies": [ { "id": 1784528, "postDate": "2022-05-11T08:21:44.933Z", "content": "Thank you !
", "rawMarkdown": "Thank you !\n" }, { "id": 1786564, "postDate": "2022-05-13T02:51:43.100Z", "content": "Too strong…
", "rawMarkdown": "Too strong..." }, { "id": 1786578, "postDate": "2022-05-13T03:28:30.210Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 1784354, "postDate": "2022-05-11T06:05:03.537Z", "content": "Just want to make sure the ensemble won't exceed the time limit.
", "rawMarkdown": "Just want to make sure the ensemble won't exceed the time limit.", "votes": 2 } ], "comments": [ { "id": 1789511, "author_name": "Zhongkai Shangguan", "author_url": "", "post_date": "2022-05-14T01:04:01.757000", "content": "Just to clarify the solution helps me reach #1 in public lb is not an ensemble of some models, the solution is pretty strange and contains some external resources under non-commercial license, not sure if the host will take it. Honestly if it is not almost the end of the competition, I will make the solution public available, but now I will publish this solution right after the competition ends. I will also make one of my final submissions contains ensemble of several models that were only trained on this competition data, which should rank around #25 currently.
", "votes": 8, "replies": [ { "id": 1789579, "author_name": "ynktk", "author_url": "", "post_date": "2022-05-14T03:05:02.483000", "content": "I suggest you to read the rules regarding external data and this thread again.
\nHosts don't allow to use data with No Derivatives licenses. For example, this is rare species in Hawaii but is not included in the training data as it is granted No Derivatives licenses. We can't use data in the Macaulay Library either.
@leonshangguan
\nFYI: note that the mentioning about external data by competition host:
\nhttps://www.kaggle.com/competitions/birdclef-2022/discussion/315361#1737178
@leonshangguan I doubt that you can back your final solution with a second submission that does not contain unapproved external data.
\nIf your best submission contains 'illegal' data, I think you'll simply get removed from the LB, you won't get down to your second best submission.
@leonshangguan ok, spit it out man, what's the weird solution. Is it template matching ? Ivector?
", "votes": 0, "replies": [] }, { "id": 1800453, "author_name": "Zhongkai Shangguan", "author_url": "", "post_date": "2022-05-25T01:37:10.013000", "content": "Working on writing a discussion bro, give me 1h. But you can see our inference notebooks here and here. Promise me don't say the \"F\" word after seeing the solution.
", "votes": 3, "replies": [] }, { "id": 1800484, "author_name": "yukiya", "author_url": "", "post_date": "2022-05-25T02:08:56.943000", "content": "haha, not at all. Thanks for satisfying my curiosity.
\nIt teaches me to do a more proper literature study. Well, next time! Congratulations by the way, awesome result !
Looks like your ensemble is doing alright 😂
\nCongrats!
approximately 5,500 recordings
\nhttps://www.kaggle.com/competitions/birdclef-2022/data
", "votes": 2, "replies": [ { "id": 1784528, "author_name": "Zhongkai Shangguan", "author_url": "", "post_date": "2022-05-11T08:21:44.933000", "content": "Thank you !
", "votes": 0, "replies": [] }, { "id": 1786564, "author_name": "UEMU", "author_url": "", "post_date": "2022-05-13T02:51:43.100000", "content": "Too strong…
", "votes": 0, "replies": [] }, { "id": 1786578, "author_name": "", "author_url": "", "post_date": "2022-05-13T03:28:30.210000", "content": "", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1789511": "Just to clarify the solution helps me reach #1 in public lb is not an ensemble of some models, the solution is pretty strange and contains some external resources under non-commercial license, not sure if the host will take it. Honestly if it is not almost the end of the competition, I will make the solution public available, but now I will publish this solution right after the competition ends. I will also make one of my final submissions contains ensemble of several models that were only trained on this competition data, which should rank around #25 currently.", "1787140": "Looks like your ensemble is doing alright 😂\nCongrats!", "1784375": "approximately 5,500 recordings\n\nhttps://www.kaggle.com/competitions/birdclef-2022/data", "1784354": "Just want to make sure the ensemble won't exceed the time limit." } }