{ "id": 177008, "title": "Submission CSV not found", "url": "/competitions/birdsong-recognition/discussion/177008", "author_name": "", "post_date": "2020-08-24T12:58:54.625425600Z", "votes": 4, "comment_count": 12, "views": 0, "content": "

After 3 days of fighting to do a submission… I got this error (image attached)

\n

Submission CSV not found

\n

Is this because of

\n
    \n
  1. memory overload (or)
    2. \n
  2. time out (or)
    3. \n
  3. some error somewhere in code
    4. \n
  4. anything else
    5. \n
\n

Will I get an error something like specifically mentioning \"memory overflow\" or \"time out\" or \"error while execution\" etc.,
\nor is it the same error \"Submission CSV not found\" for every reason

\n

Anyone who did a successful submission and came across this error. Please help me. Any other information on other error debugging during submission will help (when I come across them later).

\n

Thanks.
\nThis is my first Competition

", "messages": [ { "id": "983618", "postDate": "08/24/2020 12:58:54", "content": "

After 3 days of fighting to do a submission… I got this error (image attached)

\n

Submission CSV not found

\n

Is this because of

\n
    \n
  1. memory overload (or)
    2. \n
  2. time out (or)
    3. \n
  3. some error somewhere in code
    4. \n
  4. anything else
    5. \n
\n

Will I get an error something like specifically mentioning \"memory overflow\" or \"time out\" or \"error while execution\" etc.,
\nor is it the same error \"Submission CSV not found\" for every reason

\n

Anyone who did a successful submission and came across this error. Please help me. Any other information on other error debugging during submission will help (when I come across them later).

\n

Thanks.
\nThis is my first Competition

", "rawMarkdown": "After 3 days of fighting to do a submission... I got this error (image attached)\n\nSubmission CSV not found\n\nIs this because of \n1. memory overload (or)\n2. time out (or)\n3. some error somewhere in code\n4. anything else\n\nWill I get an error something like specifically mentioning \"memory overflow\" or \"time out\" or \"error while execution\" etc.,\nor is it the same error \"Submission CSV not found\" for every reason\n\nAnyone who did a successful submission and came across this error. Please help me. Any other information on other error debugging during submission will help (when I come across them later).\n\nThanks.\nThis is my first Competition", "votes": null }, { "id": "984007", "postDate": "08/24/2020 19:00:43", "content": "

Do you still need help? I see you submitted to the comp.

", "rawMarkdown": "Do you still need help? I see you submitted to the comp.", "votes": null }, { "id": "984038", "postDate": "08/24/2020 19:27:49", "content": "

That was a forked notebook (not my original work) and the score is sample_submission - all nocall case

\n

I am still stuck

\n

Right now I am working on this way of debugging

\n

debug =0
\ntry:
\n{
\ncodeblock
\n}
\nexcept:
\ndebug =1

\n

if debug=0:
\nsubmit model_submission
\nelse:
\nsubmit sample_submission

\n

planning to do this multiple times around my code several times…
\nThe worst part in this story is 2 submissions/day

\n

any ideas and solutions from ur side will surely help

", "rawMarkdown": "That was a forked notebook (not my original work) and the score is sample_submission - all nocall case\n\nI am still stuck\n\nRight now I am working on this way of debugging\n\ndebug =0\ntry:\n{\ncodeblock\n}\nexcept:\ndebug =1 \n\nif debug=0:\nsubmit model_submission\nelse:\nsubmit sample_submission\n\nplanning to do this multiple times around my code several times... \nThe worst part in this story is *2 submissions/day*\n\nany ideas and solutions from ur side will surely help", "votes": null }, { "id": "984060", "postDate": "08/24/2020 19:53:00", "content": "

How long does it take until that error comes up? 1 hour?

", "rawMarkdown": "How long does it take until that error comes up? 1 hour?", "votes": null }, { "id": "984103", "postDate": "08/24/2020 20:39:30", "content": "

close to 2hrs 30 mins… I timed it and recorded every 10 mins

", "rawMarkdown": "close to 2hrs 30 mins... I timed it and recorded every 10 mins", "votes": null }, { "id": "984111", "postDate": "08/24/2020 20:49:55", "content": "

Your CSV is called: submission.csv and your columns are: row_id and birds ?

", "rawMarkdown": "Your CSV is called: submission.csv and your columns are: row_id and birds ?", "votes": null }, { "id": "984115", "postDate": "08/24/2020 20:55:11", "content": "

yes… I downloaded this after commit on a fakeset

", "rawMarkdown": "yes... I downloaded this after commit on a fakeset", "votes": null }, { "id": "984575", "postDate": "08/25/2020 06:52:32", "content": "

I'm having the same problem.
\nSubmission CSV is not found after 1 hour of processing.
\nI am absolutely sure that I am writing the file 'submission.csv' in the current directory.
\nI also I don't get to see the actual output of the run, only the output of the last run of my notebook outside the competition.
\nSo there is no way for me to figure out what is going wrong.
\nAlso very frustrating is you only get 2 submissions per day, even when the submission fails.

", "rawMarkdown": "I'm having the same problem.\nSubmission CSV is not found after 1 hour of processing.\nI am absolutely sure that I am writing the file 'submission.csv' in the current directory.\nI also I don't get to see the actual output of the run, only the output of the last run of my notebook outside the competition.\nSo there is no way for me to figure out what is going wrong. \nAlso very frustrating is you only get 2 submissions per day, even when the submission fails.", "votes": null }, { "id": "984757", "postDate": "08/25/2020 09:11:04", "content": "

@vsvrp1995 Hi Ravi, are you using disk space to generate images? I am, and maybe the reason is that we're running out of disk space.

", "rawMarkdown": "vsvrp1995 Hi Ravi, are you using disk space to generate images? I am, and maybe the reason is that we're running out of disk space.", "votes": null }, { "id": "985048", "postDate": "08/25/2020 13:20:42", "content": "

Solved it…
\n@botkop

\n

Yes, I am using disk space… I am still using disk space in my succeded notebook… I did quite some changes in my code based on these points… One of these or Both of these could be the reason for failure

\n

1) I think there is an overflow of RAM… - I generate melspec and save them using matplotlib.
\n Changes regarding this -

\n\n

I have a faketest database (currently private) which has 60 recordings with total duration of 1500 mins - I created this to test my code on a large dataset (mimicking competition database)
\nWithout these changes - ram usage - 9 GB, time taken - 58 min (CPU only)
\nAfter these changes - ram usage - 4GB, time taken - 1hr 25 min (CPU only) - So time increased and ram usage decreased.

\n

2) Delete all created files (other than submission.csv files) just before submission.

\n

import shutil
\nshutil.rmtree(test_filesave_dir) # this is my folder where I saved all data on disk '/kaggle/working/test'

\n

I read this solution on discussion forums on other websites. and it worked for them and me.
\nSince u r also using diskspace… please do this

\n

Shared my faketest database (https://www.kaggle.com/vsvrp1995/faketest) - run your code on this database instead of the birdcall-check database. Observer your resource limits (disk space, RAM, time taken).

\n

The score I got is 0.543 (0.544 score is for sample_submission.csv)… So I think it submitted my actual submission file based on my model (which performed slightly worse than sample_submission.csv)

\n

There is still a small chance that It did not succeed and even this 0.543 is also because of sample_submission.csv… (just a 0.001 lower than somehow)

\n

So I will submit again and comment here (if I failed again)

\n

Currently, I am out of submissions for today… So will do it first thing tomorrow

\n

If you manage to succeed after doing one or both these changes… or any other changes… Please comment here.

", "rawMarkdown": "Solved it... \n@botkop \n\nYes, I am using disk space... I am still using disk space in my succeded notebook... I did quite some changes in my code based on these points... One of these or Both of these could be the reason for failure\n\n1) I think there is an overflow of RAM... - I generate melspec and save them using matplotlib.\n Changes regarding this - \n- using del variables - wherever i can \n- particularly del fig (where fig = plt.figure()).... - this reduced my ram usage to at least half\n- gc.collect()\n\nI have a faketest database (currently private) which has 60 recordings with total duration of 1500 mins - I created this to test my code on a large dataset (mimicking competition database)\nWithout these changes - ram usage - 9 GB, time taken - 58 min (CPU only)\nAfter these changes - ram usage - 4GB, time taken - 1hr 25 min (CPU only) - So time increased and ram usage decreased.\n\n 2) Delete all created files (other than submission.csv files) just before submission.\n\nimport shutil\nshutil.rmtree(test_filesave_dir) # this is my folder where I saved all data on disk '/kaggle/working/test'\n\nI read this solution on discussion forums on other websites. and it worked for them and me.\nSince u r also using diskspace... please do this\n\nShared my faketest database (https://www.kaggle.com/vsvrp1995/faketest) - run your code on this database instead of the birdcall-check database. Observer your resource limits (disk space, RAM, time taken).\n\nThe score I got is 0.543 (0.544 score is for sample_submission.csv)... So I think it submitted my actual submission file based on my model (which performed slightly worse than sample_submission.csv)\n\nThere is still a small chance that It did not succeed and even this 0.543 is also because of sample_submission.csv... (just a 0.001 lower than somehow)\n\nSo I will submit again and comment here (if I failed again)\n\nCurrently, I am out of submissions for today... So will do it first thing tomorrow\n\nIf you manage to succeed after doing one or both these changes... or any other changes... Please comment here.", "votes": null }, { "id": "986312", "postDate": "08/26/2020 11:16:37", "content": "

Hi Ravi,

\n

I was also able to solve the problem, by managing disk space and memory usage a bit more carefully.
\nThanks for your help.

\n

Kind regards,
\nKoen.

", "rawMarkdown": "Hi Ravi,\n\nI was also able to solve the problem, by managing disk space and memory usage a bit more carefully.\nThanks for your help.\n\nKind regards,\nKoen.", "votes": null }, { "id": "987142", "postDate": "08/27/2020 03:36:15", "content": "

Remember, Kaggle \"disk space\" is actually memory; only data files attached to the kernel do not burden your kernel memory. Any files you generate and save to the \"temp\" directory, actually consume part of your memory budget.\nThe test set may have some large files that will challenge your system, so careful book keeping is necessary. If your kernel crashes, there will be no message, no trace, just no submission generated. Predicting or processing in a try catch may help.

\n\n

p.s. Kaggle is test marketing actually providing real, e.g. not in-memory disk space, so they've gotten this complaint a lot

", "rawMarkdown": "Remember, Kaggle \"disk space\" is actually memory; only data files attached to the kernel do not burden your kernel memory. Any files you generate and save to the \"temp\" directory, actually consume part of your memory budget.\nThe test set may have some large files that will challenge your system, so careful book keeping is necessary. If your kernel crashes, there will be no message, no trace, just no submission generated. Predicting or processing in a try catch may help.\n\np.s. Kaggle is test marketing actually providing real, e.g. not in-memory disk space, so they've gotten this complaint a lot", "votes": null }, { "id": "989794", "postDate": "08/29/2020 05:45:16", "content": "

@aliabdin1 , @botkop , @vsvrp1995 , @toddcook, The Problem of getting an error of 'Found No submission CSV' is not because you have no csv, but because you have got other output files along with it.
\n use os.remove() to remove all other files other than 'submission.csv'
\nrefer this Notebook for reference: Notebook 1_Part 3
\n{and Upvote if you found it helpful}

", "rawMarkdown": "aliabdin1 , @botkop , @vsvrp1995 , @toddcook, The Problem of getting an error of 'Found No submission CSV' is not because you have no csv, but because you have got other output files along with it. \n use os.remove() to remove all other files other than 'submission.csv' \nrefer this Notebook for reference: [Notebook 1_Part 3](https://www.kaggle.com/timothyalexjohn/birdsong-classifier-keras-cnn-part-3-notebook-1)\n{and Upvote if you found it helpful}", "votes": null } ], "comments": [ { "id": 984007, "author_name": "aliabdin1", "author_url": "", "post_date": "08/24/2020 19:00:43", "content": "

Do you still need help? I see you submitted to the comp.

", "votes": null, "replies": [ { "id": 984038, "author_name": "vsvrp1995", "author_url": "", "post_date": "08/24/2020 19:27:49", "content": "

That was a forked notebook (not my original work) and the score is sample_submission - all nocall case

\n

I am still stuck

\n

Right now I am working on this way of debugging

\n

debug =0
\ntry:
\n{
\ncodeblock
\n}
\nexcept:
\ndebug =1

\n

if debug=0:
\nsubmit model_submission
\nelse:
\nsubmit sample_submission

\n

planning to do this multiple times around my code several times…
\nThe worst part in this story is 2 submissions/day

\n

any ideas and solutions from ur side will surely help

", "votes": null, "replies": [] }, { "id": 984060, "author_name": "aliabdin1", "author_url": "", "post_date": "08/24/2020 19:53:00", "content": "

How long does it take until that error comes up? 1 hour?

", "votes": null, "replies": [] }, { "id": 984103, "author_name": "vsvrp1995", "author_url": "", "post_date": "08/24/2020 20:39:30", "content": "

close to 2hrs 30 mins… I timed it and recorded every 10 mins

", "votes": null, "replies": [] }, { "id": 984111, "author_name": "aliabdin1", "author_url": "", "post_date": "08/24/2020 20:49:55", "content": "

Your CSV is called: submission.csv and your columns are: row_id and birds ?

", "votes": null, "replies": [] }, { "id": 984115, "author_name": "vsvrp1995", "author_url": "", "post_date": "08/24/2020 20:55:11", "content": "

yes… I downloaded this after commit on a fakeset

", "votes": null, "replies": [] } ] }, { "id": 984575, "author_name": "botkop", "author_url": "", "post_date": "08/25/2020 06:52:32", "content": "

I'm having the same problem.
\nSubmission CSV is not found after 1 hour of processing.
\nI am absolutely sure that I am writing the file 'submission.csv' in the current directory.
\nI also I don't get to see the actual output of the run, only the output of the last run of my notebook outside the competition.
\nSo there is no way for me to figure out what is going wrong.
\nAlso very frustrating is you only get 2 submissions per day, even when the submission fails.

", "votes": null, "replies": [] }, { "id": 984757, "author_name": "botkop", "author_url": "", "post_date": "08/25/2020 09:11:04", "content": "

@vsvrp1995 Hi Ravi, are you using disk space to generate images? I am, and maybe the reason is that we're running out of disk space.

", "votes": null, "replies": [] }, { "id": 985048, "author_name": "vsvrp1995", "author_url": "", "post_date": "08/25/2020 13:20:42", "content": "

Solved it…
\n@botkop

\n

Yes, I am using disk space… I am still using disk space in my succeded notebook… I did quite some changes in my code based on these points… One of these or Both of these could be the reason for failure

\n

1) I think there is an overflow of RAM… - I generate melspec and save them using matplotlib.
\n Changes regarding this -

\n\n

I have a faketest database (currently private) which has 60 recordings with total duration of 1500 mins - I created this to test my code on a large dataset (mimicking competition database)
\nWithout these changes - ram usage - 9 GB, time taken - 58 min (CPU only)
\nAfter these changes - ram usage - 4GB, time taken - 1hr 25 min (CPU only) - So time increased and ram usage decreased.

\n

2) Delete all created files (other than submission.csv files) just before submission.

\n

import shutil
\nshutil.rmtree(test_filesave_dir) # this is my folder where I saved all data on disk '/kaggle/working/test'

\n

I read this solution on discussion forums on other websites. and it worked for them and me.
\nSince u r also using diskspace… please do this

\n

Shared my faketest database (https://www.kaggle.com/vsvrp1995/faketest) - run your code on this database instead of the birdcall-check database. Observer your resource limits (disk space, RAM, time taken).

\n

The score I got is 0.543 (0.544 score is for sample_submission.csv)… So I think it submitted my actual submission file based on my model (which performed slightly worse than sample_submission.csv)

\n

There is still a small chance that It did not succeed and even this 0.543 is also because of sample_submission.csv… (just a 0.001 lower than somehow)

\n

So I will submit again and comment here (if I failed again)

\n

Currently, I am out of submissions for today… So will do it first thing tomorrow

\n

If you manage to succeed after doing one or both these changes… or any other changes… Please comment here.

", "votes": null, "replies": [] }, { "id": 986312, "author_name": "botkop", "author_url": "", "post_date": "08/26/2020 11:16:37", "content": "

Hi Ravi,

\n

I was also able to solve the problem, by managing disk space and memory usage a bit more carefully.
\nThanks for your help.

\n

Kind regards,
\nKoen.

", "votes": null, "replies": [] }, { "id": 989794, "author_name": "timothyalexjohn", "author_url": "", "post_date": "08/29/2020 05:45:16", "content": "

@aliabdin1 , @botkop , @vsvrp1995 , @toddcook, The Problem of getting an error of 'Found No submission CSV' is not because you have no csv, but because you have got other output files along with it.
\n use os.remove() to remove all other files other than 'submission.csv'
\nrefer this Notebook for reference: Notebook 1_Part 3
\n{and Upvote if you found it helpful}

", "votes": null, "replies": [] }, { "id": 987142, "author_name": "toddcook", "author_url": "", "post_date": "08/27/2020 03:36:15", "content": "

Remember, Kaggle \"disk space\" is actually memory; only data files attached to the kernel do not burden your kernel memory. Any files you generate and save to the \"temp\" directory, actually consume part of your memory budget.\nThe test set may have some large files that will challenge your system, so careful book keeping is necessary. If your kernel crashes, there will be no message, no trace, just no submission generated. Predicting or processing in a try catch may help.

\n\n

p.s. Kaggle is test marketing actually providing real, e.g. not in-memory disk space, so they've gotten this complaint a lot

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "983618": "After 3 days of fighting to do a submission... I got this error (image attached)\n\nSubmission CSV not found\n\nIs this because of \n1. memory overload (or)\n2. time out (or)\n3. some error somewhere in code\n4. anything else\n\nWill I get an error something like specifically mentioning \"memory overflow\" or \"time out\" or \"error while execution\" etc.,\nor is it the same error \"Submission CSV not found\" for every reason\n\nAnyone who did a successful submission and came across this error. Please help me. Any other information on other error debugging during submission will help (when I come across them later).\n\nThanks.\nThis is my first Competition", "984007": "Do you still need help? I see you submitted to the comp.", "984038": "That was a forked notebook (not my original work) and the score is sample_submission - all nocall case\n\nI am still stuck\n\nRight now I am working on this way of debugging\n\ndebug =0\ntry:\n{\ncodeblock\n}\nexcept:\ndebug =1 \n\nif debug=0:\nsubmit model_submission\nelse:\nsubmit sample_submission\n\nplanning to do this multiple times around my code several times... \nThe worst part in this story is *2 submissions/day*\n\nany ideas and solutions from ur side will surely help", "984060": "How long does it take until that error comes up? 1 hour?", "984103": "close to 2hrs 30 mins... I timed it and recorded every 10 mins", "984111": "Your CSV is called: submission.csv and your columns are: row_id and birds ?", "984115": "yes... I downloaded this after commit on a fakeset", "984575": "I'm having the same problem.\nSubmission CSV is not found after 1 hour of processing.\nI am absolutely sure that I am writing the file 'submission.csv' in the current directory.\nI also I don't get to see the actual output of the run, only the output of the last run of my notebook outside the competition.\nSo there is no way for me to figure out what is going wrong. \nAlso very frustrating is you only get 2 submissions per day, even when the submission fails.", "984757": "vsvrp1995 Hi Ravi, are you using disk space to generate images? I am, and maybe the reason is that we're running out of disk space.", "985048": "Solved it... \n@botkop \n\nYes, I am using disk space... I am still using disk space in my succeded notebook... I did quite some changes in my code based on these points... One of these or Both of these could be the reason for failure\n\n1) I think there is an overflow of RAM... - I generate melspec and save them using matplotlib.\n Changes regarding this - \n- using del variables - wherever i can \n- particularly del fig (where fig = plt.figure()).... - this reduced my ram usage to at least half\n- gc.collect()\n\nI have a faketest database (currently private) which has 60 recordings with total duration of 1500 mins - I created this to test my code on a large dataset (mimicking competition database)\nWithout these changes - ram usage - 9 GB, time taken - 58 min (CPU only)\nAfter these changes - ram usage - 4GB, time taken - 1hr 25 min (CPU only) - So time increased and ram usage decreased.\n\n 2) Delete all created files (other than submission.csv files) just before submission.\n\nimport shutil\nshutil.rmtree(test_filesave_dir) # this is my folder where I saved all data on disk '/kaggle/working/test'\n\nI read this solution on discussion forums on other websites. and it worked for them and me.\nSince u r also using diskspace... please do this\n\nShared my faketest database (https://www.kaggle.com/vsvrp1995/faketest) - run your code on this database instead of the birdcall-check database. Observer your resource limits (disk space, RAM, time taken).\n\nThe score I got is 0.543 (0.544 score is for sample_submission.csv)... So I think it submitted my actual submission file based on my model (which performed slightly worse than sample_submission.csv)\n\nThere is still a small chance that It did not succeed and even this 0.543 is also because of sample_submission.csv... (just a 0.001 lower than somehow)\n\nSo I will submit again and comment here (if I failed again)\n\nCurrently, I am out of submissions for today... So will do it first thing tomorrow\n\nIf you manage to succeed after doing one or both these changes... or any other changes... Please comment here.", "986312": "Hi Ravi,\n\nI was also able to solve the problem, by managing disk space and memory usage a bit more carefully.\nThanks for your help.\n\nKind regards,\nKoen.", "987142": "Remember, Kaggle \"disk space\" is actually memory; only data files attached to the kernel do not burden your kernel memory. Any files you generate and save to the \"temp\" directory, actually consume part of your memory budget.\nThe test set may have some large files that will challenge your system, so careful book keeping is necessary. If your kernel crashes, there will be no message, no trace, just no submission generated. Predicting or processing in a try catch may help.\n\np.s. Kaggle is test marketing actually providing real, e.g. not in-memory disk space, so they've gotten this complaint a lot", "989794": "aliabdin1 , @botkop , @vsvrp1995 , @toddcook, The Problem of getting an error of 'Found No submission CSV' is not because you have no csv, but because you have got other output files along with it. \n use os.remove() to remove all other files other than 'submission.csv' \nrefer this Notebook for reference: [Notebook 1_Part 3](https://www.kaggle.com/timothyalexjohn/birdsong-classifier-keras-cnn-part-3-notebook-1)\n{and Upvote if you found it helpful}" }, "source": "meta" }