![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2972565%2F68d47b88ea48e23aa5f5271d88293b14%2F1.jpg?generation=1582342847805088&alt=media\")
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2972565%2F556252e680ccbb8d8ef19e27f4cd5cbd%2F3.jpg?generation=1582343071674677&alt=media\")
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2972565%2Fe9e0f16ffe90064f0f556118cec98728%2F4.jpg?generation=1582343712060328&alt=media\")
why it is happened? i didnt get score until now..
\n\nplease give me a adivice.. or kaggle team please solve my error
\n\n\n\n
why it is happened? i didnt get score until now..
\n\nplease give me a adivice.. or kaggle team please solve my error
\n\n\n\n
Can you try to sort the order of files A-Z in submission.csv, I believe that is how they are sorted in Sample Submission.\nThen Submit your file
", "rawMarkdown": "Can you try to sort the order of files A-Z in submission.csv, I believe that is how they are sorted in Sample Submission.\nThen Submit your file", "votes": null }, { "id": "753420", "postDate": "02/22/2020 06:35:58", "content": "Sorting isn't required according to me. My submission got accepted in which the filenames weren't sorted. But I did face the same issue earlier for which I've explained what to do. I think that should work.
", "rawMarkdown": "Sorting isn't required according to me. My submission got accepted in which the filenames weren't sorted. But I did face the same issue earlier for which I've explained what to do. I think that should work.", "votes": null }, { "id": "753424", "postDate": "02/22/2020 06:42:11", "content": "No Harm in trying and staying organized ;)
", "rawMarkdown": "No Harm in trying and staying organized ;)", "votes": null }, { "id": "753425", "postDate": "02/22/2020 06:45:31", "content": "Sure.
", "rawMarkdown": "Sure.", "votes": null }, { "id": "753443", "postDate": "02/22/2020 07:30:19", "content": "how to sure the external dataset? \ni checked that internet is disabled.\nif u have a example for submission can u share to me ? \ni think my submission maker contain all file that is why i filter non .mp4 files.. \nis it right thinking?
", "rawMarkdown": "how to sure the external dataset? \ni checked that internet is disabled.\nif u have a example for submission can u share to me ? \ni think my submission maker contain all file that is why i filter non .mp4 files.. \nis it right thinking?", "votes": null }, { "id": "753536", "postDate": "02/22/2020 10:30:08", "content": "External datasets don't need to be public. This it not the issue.
", "rawMarkdown": "External datasets don't need to be public. This it not the issue.", "votes": null }, { "id": "753538", "postDate": "02/22/2020 10:33:48", "content": "This error means your notebook/script fails before you write submission.csv.\nThis failure can have many reasons. Usually, it's because you hard coded something, or maybe reading one of the first files failed, etc.\nYou can find here how I was able to debug such issues in another competition. If you'll look around you'll find more discussions about it
", "rawMarkdown": "This error means your notebook/script fails before you write submission.csv.\nThis failure can have many reasons. Usually, it's because you hard coded something, or maybe reading one of the first files failed, etc.\nYou can find [here](https://www.kaggle.com/c/google-quest-challenge/discussion/120368) how I was able to debug such issues in another competition. If you'll look around you'll find more discussions about it", "votes": null }, { "id": "753705", "postDate": "02/22/2020 15:15:21", "content": "It's mentioned external data which includes the model needs to be public.
", "rawMarkdown": "It's mentioned external data which includes the model needs to be public.", "votes": null }, { "id": "753706", "postDate": "02/22/2020 15:17:48", "content": "Go to this link. On the top-right corner, you can see \"New Dataset\". Using this you can upload your externally trained model (and make that public). Later when you open your notebook, you will find an option to \"Add Data\". Use that to add your uploaded model.
", "rawMarkdown": "Go to this [link](https://www.kaggle.com/datasets). On the top-right corner, you can see \"New Dataset\". Using this you can upload your externally trained model (and make that public). Later when you open your notebook, you will find an option to \"Add Data\". Use that to add your uploaded model.", "votes": null }, { "id": "753709", "postDate": "02/22/2020 15:26:11", "content": "i also use mtcnn as external data do i need to make public???\ni confused about that.. \nand i will try to submit with changing file filter on tomorrow..\nif u guys have any advice.. please give me advice
", "rawMarkdown": "i also use mtcnn as external data do i need to make public???\ni confused about that.. \nand i will try to submit with changing file filter on tomorrow..\nif u guys have any advice.. please give me advice", "votes": null }, { "id": "753720", "postDate": "02/22/2020 15:42:49", "content": "what, is this true? But if I make my weights public other people can easily beat my score?
", "rawMarkdown": "what, is this true? But if I make my weights public other people can easily beat my score?", "votes": null }, { "id": "753734", "postDate": "02/22/2020 15:52:58", "content": "\n\n\nBut if I make my weights public
\n
When you create a new dataset it is not automatically public. So just don't make it public. :-)
", "rawMarkdown": "> But if I make my weights public\n\nWhen you create a new dataset it is not automatically public. So just don't make it public. :-)", "votes": null }, { "id": "753737", "postDate": "02/22/2020 15:56:36", "content": "@saanikagupta The phrase \"External data\" is a little confusing here. It refers to data you didn't create yourself. You can (and should) train your models and load their weights as \"Private\" datasets (otherwise you would be able to use the model trained by the leading teams). But when you use data you didn't create - like pre-trained models, etc., you should make this external data public. \nThe kernel itself won't fail if you didn't make this data public, but at the end of the competition you might me disqualified.
", "rawMarkdown": "saanikagupta The phrase \"External data\" is a little confusing here. It refers to data you didn't create yourself. You can (and should) train your models and load their weights as \"Private\" datasets (otherwise you would be able to use the model trained by the leading teams). But when you use data you didn't create - like pre-trained models, etc., you should make this external data public. \nThe kernel itself won't fail if you didn't make this data public, but at the end of the competition you might me disqualified.", "votes": null }, { "id": "753763", "postDate": "02/22/2020 16:24:05", "content": "should i make public just pre train model weight file?
\n\nor
\n\nmake public all?
\n\nand how to make mydata public?
", "rawMarkdown": "should i make public just pre train model weight file? \n\nor\n\nmake public all?\n\nand how to make mydata public?", "votes": null }, { "id": "753789", "postDate": "02/22/2020 16:45:03", "content": "Hi Choo,
\n\nLooking at the screenshot you shared of your submission, for image #3 in your submission file the probability is 1.0 when you input such whole number as probability inot Log Loss function,it can throw an error as there is no log of 0, hence there is a minmax rule for LogLoss please check this wikipedia of fast.ai http://wiki.fast.ai/index.php/Log_Loss and scroll down to MinMax Rule. Try to change that and submit.
\n\nThanks\nSunny
", "rawMarkdown": "Hi Choo,\n\nLooking at the screenshot you shared of your submission, for image #3 in your submission file the probability is 1.0 when you input such whole number as probability inot Log Loss function,it can throw an error as there is no log of 0, hence there is a minmax rule for LogLoss please check this wikipedia of fast.ai [http://wiki.fast.ai/index.php/Log_Loss](http://wiki.fast.ai/index.php/Log_Loss) and scroll down to MinMax Rule. Try to change that and submit.\n\nThanks\nSunny", "votes": null }, { "id": "753803", "postDate": "02/22/2020 16:56:32", "content": "\n\n\nit can throw an error as there is no log of 0
\n
Kaggle will already do this for you (it adds an epsilon of 1e-15), so predicting 1 or 0 is fine.
", "rawMarkdown": "> it can throw an error as there is no log of 0\n\nKaggle will already do this for you (it adds an epsilon of 1e-15), so predicting 1 or 0 is fine.", "votes": null }, { "id": "753810", "postDate": "02/22/2020 17:00:49", "content": "Yes Just defined a logloss function with epsilon of 1e-15 and it does return some value.
", "rawMarkdown": "Yes Just defined a logloss function with epsilon of 1e-15 and it does return some value.", "votes": null }, { "id": "753883", "postDate": "02/22/2020 19:19:27", "content": "You don't have to make anything public for the notebook not to fail. If you use an external dataset (even a public one) just post it in the \"External Data\" thread
", "rawMarkdown": "You don't have to make anything public for the notebook not to fail. If you use an external dataset (even a public one) just post it in the \"External Data\" thread", "votes": null }, { "id": "753996", "postDate": "02/22/2020 22:59:00", "content": "i got it thank........ anyway i try after 1 hour again\ni hope to submit submission file successfully...
", "rawMarkdown": "i got it thank........ anyway i try after 1 hour again\ni hope to submit submission file successfully...", "votes": null } ], "comments": [ { "id": 753368, "author_name": "saanikagupta", "author_url": "", "post_date": "02/22/2020 05:03:41", "content": "", "votes": null, "replies": [ { "id": 753443, "author_name": "chooyoungjun", "author_url": "", "post_date": "02/22/2020 07:30:19", "content": "how to sure the external dataset? \ni checked that internet is disabled.\nif u have a example for submission can u share to me ? \ni think my submission maker contain all file that is why i filter non .mp4 files.. \nis it right thinking?
", "votes": null, "replies": [] }, { "id": 753536, "author_name": "yuval6967", "author_url": "", "post_date": "02/22/2020 10:30:08", "content": "External datasets don't need to be public. This it not the issue.
", "votes": null, "replies": [] }, { "id": 753705, "author_name": "saanikagupta", "author_url": "", "post_date": "02/22/2020 15:15:21", "content": "It's mentioned external data which includes the model needs to be public.
", "votes": null, "replies": [] }, { "id": 753709, "author_name": "chooyoungjun", "author_url": "", "post_date": "02/22/2020 15:26:11", "content": "i also use mtcnn as external data do i need to make public???\ni confused about that.. \nand i will try to submit with changing file filter on tomorrow..\nif u guys have any advice.. please give me advice
", "votes": null, "replies": [] }, { "id": 753720, "author_name": "guesejustin", "author_url": "", "post_date": "02/22/2020 15:42:49", "content": "what, is this true? But if I make my weights public other people can easily beat my score?
", "votes": null, "replies": [] }, { "id": 753734, "author_name": "humananalog", "author_url": "", "post_date": "02/22/2020 15:52:58", "content": "\n\n\nBut if I make my weights public
\n
When you create a new dataset it is not automatically public. So just don't make it public. :-)
", "votes": null, "replies": [] }, { "id": 753737, "author_name": "yuval6967", "author_url": "", "post_date": "02/22/2020 15:56:36", "content": "@saanikagupta The phrase \"External data\" is a little confusing here. It refers to data you didn't create yourself. You can (and should) train your models and load their weights as \"Private\" datasets (otherwise you would be able to use the model trained by the leading teams). But when you use data you didn't create - like pre-trained models, etc., you should make this external data public. \nThe kernel itself won't fail if you didn't make this data public, but at the end of the competition you might me disqualified.
", "votes": null, "replies": [] }, { "id": 753763, "author_name": "chooyoungjun", "author_url": "", "post_date": "02/22/2020 16:24:05", "content": "should i make public just pre train model weight file?
\n\nor
\n\nmake public all?
\n\nand how to make mydata public?
", "votes": null, "replies": [] }, { "id": 753883, "author_name": "yuval6967", "author_url": "", "post_date": "02/22/2020 19:19:27", "content": "You don't have to make anything public for the notebook not to fail. If you use an external dataset (even a public one) just post it in the \"External Data\" thread
", "votes": null, "replies": [] }, { "id": 753996, "author_name": "chooyoungjun", "author_url": "", "post_date": "02/22/2020 22:59:00", "content": "i got it thank........ anyway i try after 1 hour again\ni hope to submit submission file successfully...
", "votes": null, "replies": [] } ] }, { "id": 753417, "author_name": "sunnykm", "author_url": "", "post_date": "02/22/2020 06:33:36", "content": "Can you try to sort the order of files A-Z in submission.csv, I believe that is how they are sorted in Sample Submission.\nThen Submit your file
", "votes": null, "replies": [ { "id": 753420, "author_name": "saanikagupta", "author_url": "", "post_date": "02/22/2020 06:35:58", "content": "Sorting isn't required according to me. My submission got accepted in which the filenames weren't sorted. But I did face the same issue earlier for which I've explained what to do. I think that should work.
", "votes": null, "replies": [] }, { "id": 753424, "author_name": "sunnykm", "author_url": "", "post_date": "02/22/2020 06:42:11", "content": "No Harm in trying and staying organized ;)
", "votes": null, "replies": [] }, { "id": 753425, "author_name": "saanikagupta", "author_url": "", "post_date": "02/22/2020 06:45:31", "content": "Sure.
", "votes": null, "replies": [] } ] }, { "id": 753538, "author_name": "yuval6967", "author_url": "", "post_date": "02/22/2020 10:33:48", "content": "This error means your notebook/script fails before you write submission.csv.\nThis failure can have many reasons. Usually, it's because you hard coded something, or maybe reading one of the first files failed, etc.\nYou can find here how I was able to debug such issues in another competition. If you'll look around you'll find more discussions about it
", "votes": null, "replies": [] }, { "id": 753706, "author_name": "saanikagupta", "author_url": "", "post_date": "02/22/2020 15:17:48", "content": "Go to this link. On the top-right corner, you can see \"New Dataset\". Using this you can upload your externally trained model (and make that public). Later when you open your notebook, you will find an option to \"Add Data\". Use that to add your uploaded model.
", "votes": null, "replies": [] }, { "id": 753789, "author_name": "sunnykm", "author_url": "", "post_date": "02/22/2020 16:45:03", "content": "Hi Choo,
\n\nLooking at the screenshot you shared of your submission, for image #3 in your submission file the probability is 1.0 when you input such whole number as probability inot Log Loss function,it can throw an error as there is no log of 0, hence there is a minmax rule for LogLoss please check this wikipedia of fast.ai http://wiki.fast.ai/index.php/Log_Loss and scroll down to MinMax Rule. Try to change that and submit.
\n\nThanks\nSunny
", "votes": null, "replies": [ { "id": 753803, "author_name": "humananalog", "author_url": "", "post_date": "02/22/2020 16:56:32", "content": "\n\n\nit can throw an error as there is no log of 0
\n
Kaggle will already do this for you (it adds an epsilon of 1e-15), so predicting 1 or 0 is fine.
", "votes": null, "replies": [] }, { "id": 753810, "author_name": "sunnykm", "author_url": "", "post_date": "02/22/2020 17:00:49", "content": "Yes Just defined a logloss function with epsilon of 1e-15 and it does return some value.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "753335": "why it is happened? i didnt get score until now.. \n\nplease give me a adivice.. or kaggle team please solve my error\n\n\n\n", "753368": "", "753417": "Can you try to sort the order of files A-Z in submission.csv, I believe that is how they are sorted in Sample Submission.\nThen Submit your file", "753420": "Sorting isn't required according to me. My submission got accepted in which the filenames weren't sorted. But I did face the same issue earlier for which I've explained what to do. I think that should work.", "753424": "No Harm in trying and staying organized ;)", "753425": "Sure.", "753443": "how to sure the external dataset? \ni checked that internet is disabled.\nif u have a example for submission can u share to me ? \ni think my submission maker contain all file that is why i filter non .mp4 files.. \nis it right thinking?", "753536": "External datasets don't need to be public. This it not the issue.", "753538": "This error means your notebook/script fails before you write submission.csv.\nThis failure can have many reasons. Usually, it's because you hard coded something, or maybe reading one of the first files failed, etc.\nYou can find [here](https://www.kaggle.com/c/google-quest-challenge/discussion/120368) how I was able to debug such issues in another competition. If you'll look around you'll find more discussions about it", "753705": "It's mentioned external data which includes the model needs to be public.", "753706": "Go to this [link](https://www.kaggle.com/datasets). On the top-right corner, you can see \"New Dataset\". Using this you can upload your externally trained model (and make that public). Later when you open your notebook, you will find an option to \"Add Data\". Use that to add your uploaded model.", "753709": "i also use mtcnn as external data do i need to make public???\ni confused about that.. \nand i will try to submit with changing file filter on tomorrow..\nif u guys have any advice.. please give me advice", "753720": "what, is this true? But if I make my weights public other people can easily beat my score?", "753734": "> But if I make my weights public\n\nWhen you create a new dataset it is not automatically public. So just don't make it public. :-)", "753737": "saanikagupta The phrase \"External data\" is a little confusing here. It refers to data you didn't create yourself. You can (and should) train your models and load their weights as \"Private\" datasets (otherwise you would be able to use the model trained by the leading teams). But when you use data you didn't create - like pre-trained models, etc., you should make this external data public. \nThe kernel itself won't fail if you didn't make this data public, but at the end of the competition you might me disqualified.", "753763": "should i make public just pre train model weight file? \n\nor\n\nmake public all?\n\nand how to make mydata public?", "753789": "Hi Choo,\n\nLooking at the screenshot you shared of your submission, for image #3 in your submission file the probability is 1.0 when you input such whole number as probability inot Log Loss function,it can throw an error as there is no log of 0, hence there is a minmax rule for LogLoss please check this wikipedia of fast.ai [http://wiki.fast.ai/index.php/Log_Loss](http://wiki.fast.ai/index.php/Log_Loss) and scroll down to MinMax Rule. Try to change that and submit.\n\nThanks\nSunny", "753803": "> it can throw an error as there is no log of 0\n\nKaggle will already do this for you (it adds an epsilon of 1e-15), so predicting 1 or 0 is fine.", "753810": "Yes Just defined a logloss function with epsilon of 1e-15 and it does return some value.", "753883": "You don't have to make anything public for the notebook not to fail. If you use an external dataset (even a public one) just post it in the \"External Data\" thread", "753996": "i got it thank........ anyway i try after 1 hour again\ni hope to submit submission file successfully..." }, "source": "meta" }