![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F654629%2Ff72b28d44001280e8705d217c3ff13b0%2FSelection_026.png?generation=1562150549397394&alt=media\")
Dear All,\nMy kernel build successfully and the respective CSV is also created with the requested name:\n
However time after time following submission, I immediately get this error:\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F654629%2Fccafbb6c48f405899a6fb3f90ef15197%2FSelection_027.png?generation=1562150597174853&alt=media\")
Any ideas as to what may be wrong? \nThanks.
", "messages": [ { "id": 567317, "postDate": "2019-07-03T10:43:49.110Z", "content": "Dear All,\nMy kernel build successfully and the respective CSV is also created with the requested name:\n
However time after time following submission, I immediately get this error:\n
Any ideas as to what may be wrong? \nThanks.
", "rawMarkdown": "Dear All,\nMy kernel build successfully and the respective CSV is also created with the requested name:\n\n\nHowever time after time following submission, I immediately get this error:\n\n\nAny ideas as to what may be wrong? \nThanks. \n ", "votes": 9 }, { "id": 569982, "postDate": "2019-07-07T16:26:52.983Z", "content": "@solomonk I am encountering a similar problem. I have TTA implemented and my notebook outputs the submission CSV file after committing. However, the status eventually displays \"Submission CSV Not Found\".
\n\nI've tweaked the batch size in predict_generator and I can't get my kernel with TTA to work.
", "rawMarkdown": "@solomonk I am encountering a similar problem. I have TTA implemented and my notebook outputs the submission CSV file after committing. However, the status eventually displays \"Submission CSV Not Found\".\n\nI've tweaked the batch size in predict_generator and I can't get my kernel with TTA to work.", "votes": 1, "replies": [ { "id": 570053, "postDate": "2019-07-07T18:03:21.807Z", "content": "For now, and since there was no response from the admins, I am not using TTA.
", "rawMarkdown": "For now, and since there was no response from the admins, I am not using TTA. ", "votes": 1 } ] }, { "id": 569939, "postDate": "2019-07-07T15:16:15.833Z", "content": "Once I removed the TTA I was able to submit. I can not fathom why this may impose a problem.
", "rawMarkdown": "Once I removed the TTA I was able to submit. I can not fathom why this may impose a problem. ", "votes": 2, "replies": [ { "id": 570381, "postDate": "2019-07-08T07:56:45.787Z", "content": "Did you also reduce batch size or image size? Because I have the same CSV not found and I trying everything with no luck! I have the lazy TTA from Abhishek's Kernel but also without TTA isn't working.\nI m truly considering not investing more time on this competition...
", "rawMarkdown": "Did you also reduce batch size or image size? Because I have the same CSV not found and I trying everything with no luck! I have the lazy TTA from Abhishek's Kernel but also without TTA isn't working.\nI m truly considering not investing more time on this competition...", "votes": 1 }, { "id": 570615, "postDate": "2019-07-08T14:43:54.740Z", "content": "I can privately share a notebook that works without TTA.
", "rawMarkdown": "I can privately share a notebook that works without TTA. " }, { "id": 570636, "postDate": "2019-07-08T15:08:09.803Z", "content": "Pretty sure we can not share private code during the competition :smile:\nBut I would love to see a kernel with that (even a low score).
", "rawMarkdown": "Pretty sure we can not share private code during the competition :smile:\nBut I would love to see a kernel with that (even a low score).", "votes": 1 } ] }, { "id": 569297, "postDate": "2019-07-06T11:38:51.690Z", "content": "I am getting the same error too. My kernel is an inference version. Only takes a few minutes to run. I started two days ago for me. It started happening when I went to larger images, but reducing image size does not fix the issue.\nFixed it. The images had to be the same size I changed to 299 from 150, but not it has to be the original size to work, not 100% why yet.
", "rawMarkdown": "I am getting the same error too. My kernel is an inference version. Only takes a few minutes to run. I started two days ago for me. It started happening when I went to larger images, but reducing image size does not fix the issue.\nFixed it. The images had to be the same size I changed to 299 from 150, but not it has to be the original size to work, not 100% why yet.", "votes": 2 }, { "id": 568506, "postDate": "2019-07-05T03:12:02.667Z", "content": "I've faced same issue. I think its memory overflow. You can avoid it by reducing batch size (keeping same resolution) or reducing resolution (keeping same batch size).
", "rawMarkdown": "I've faced same issue. I think its memory overflow. You can avoid it by reducing batch size (keeping same resolution) or reducing resolution (keeping same batch size).", "votes": 2, "replies": [ { "id": 721014, "postDate": "2020-01-17T00:05:33.493Z", "content": "I will try this method. I encountered the same issue in DFDC. There are 400 test videos and 4000 public test videos. Maybe memory is overflow.
", "rawMarkdown": "I will try this method. I encountered the same issue in DFDC. There are 400 test videos and 4000 public test videos. Maybe memory is overflow. ", "isDeleted": true } ] }, { "id": 567842, "postDate": "2019-07-04T04:37:43.840Z", "content": "I'm getting this error when using larger image sizes and when using more training images. Therefore, I suspect that this is actually a memory related error. It could be that the system is running out of RAM during prediction. Reducing the image size may help.
", "rawMarkdown": "I'm getting this error when using larger image sizes and when using more training images. Therefore, I suspect that this is actually a memory related error. It could be that the system is running out of RAM during prediction. Reducing the image size may help.", "votes": 2 }, { "id": 635261, "postDate": "2019-09-27T10:18:44.790Z", "content": "I have the same issue with other dataset . I attached the noteboook
\n\nhttps://www.kaggle.com/questions-and-answers/110395
", "rawMarkdown": "I have the same issue with other dataset . I attached the noteboook\n\nhttps://www.kaggle.com/questions-and-answers/110395\n\n" }, { "id": 615317, "postDate": "2019-09-01T18:59:50.683Z", "content": "I am also facing the same issue.If anyone has find the solution , Please share.
", "rawMarkdown": "I am also facing the same issue.If anyone has find the solution , Please share." }, { "id": 610069, "postDate": "2019-08-28T11:34:28.180Z", "content": "I'm having the same problem. My kernel runs in 2 minutes with GPU and 10 with CPU. The commit goes fine, producing the CSV in output. My code only loads the weights from a pre-trained model and predict test images. During execution, it consumes not more than 5GB of RAM.
", "rawMarkdown": "I'm having the same problem. My kernel runs in 2 minutes with GPU and 10 with CPU. The commit goes fine, producing the CSV in output. My code only loads the weights from a pre-trained model and predict test images. During execution, it consumes not more than 5GB of RAM." }, { "id": 574831, "postDate": "2019-07-14T14:42:51.193Z", "content": "Hi,
\n\nI got the same problem. The solution was for me to remove the file operations from the kernel.\nI used to save the preprocessed images first, before prediction.\nWhen I predicted the images on the fly, then I managed to submit the kernel.
\n\nSomething was wrong with saving the images to /tmp/
", "rawMarkdown": "Hi,\n\nI got the same problem. The solution was for me to remove the file operations from the kernel.\nI used to save the preprocessed images first, before prediction.\nWhen I predicted the images on the fly, then I managed to submit the kernel.\n\nSomething was wrong with saving the images to /tmp/" }, { "id": 574190, "postDate": "2019-07-13T11:02:01.473Z", "content": "I had the same problem. \nI managed to solve it by removing all the file operations.
\n\nI used to save preprocessed images. Now I preprocess and predict everything on the fly, without saving them first to the disk. It helped...
", "rawMarkdown": "I had the same problem. \nI managed to solve it by removing all the file operations.\n\nI used to save preprocessed images. Now I preprocess and predict everything on the fly, without saving them first to the disk. It helped..." }, { "id": 572077, "postDate": "2019-07-10T12:19:44.987Z", "content": "I have the same issue. Is anyone solved it?
", "rawMarkdown": "I have the same issue. Is anyone solved it?", "replies": [ { "id": 572082, "postDate": "2019-07-10T12:39:40.953Z", "content": "Yes, it should be a memory issue. In my case, it was either the image size or batch size during the training phase. TTA was not the cause.\nTry to reduce both and hopefully, it will work xD
", "rawMarkdown": "Yes, it should be a memory issue. In my case, it was either the image size or batch size during the training phase. TTA was not the cause.\nTry to reduce both and hopefully, it will work xD" } ] }, { "id": 569769, "postDate": "2019-07-07T09:35:56.177Z", "content": "I too facing this problem...Please someone help me
", "rawMarkdown": "I too facing this problem...Please someone help me" }, { "id": 567743, "postDate": "2019-07-04T00:40:08.407Z", "content": "I have the same problem
", "rawMarkdown": "I have the same problem" }, { "id": 567504, "postDate": "2019-07-03T15:52:00.720Z", "content": "I have the same problem, any ideas about how to solve it?
", "rawMarkdown": "I have the same problem, any ideas about how to solve it?", "replies": [ { "id": 567553, "postDate": "2019-07-03T17:16:49.260Z", "content": "I am trying to, once I have a solution I shall let you know.
", "rawMarkdown": "I am trying to, once I have a solution I shall let you know. " } ] }, { "id": 567489, "postDate": "2019-07-03T15:34:34.723Z", "content": "Admins, kindly let me know the exact error or I shall have to quit the competition. No kidding.
", "rawMarkdown": "Admins, kindly let me know the exact error or I shall have to quit the competition. No kidding. " }, { "id": 567343, "postDate": "2019-07-03T11:16:37.720Z", "content": "How can it be running for two hours, if the report stated that the submission file is not found ...?! \n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F654629%2Fdb54a9dab9eeee191cc76da20492aa6e%2FSelection_028.png?generation=1562152564223118&alt=media\")
The same for me. I've tried different solutions written in other discussions but no results. I think that there should be some more specific info about error (lack of resources or anything else). It is a little frustrating to change code without knowing what went wrong.
", "rawMarkdown": "The same for me. I've tried different solutions written in other discussions but no results. I think that there should be some more specific info about error (lack of resources or anything else). It is a little frustrating to change code without knowing what went wrong. ", "votes": 2 }, { "id": 567490, "postDate": "2019-07-03T15:35:01.343Z", "content": "Were you able to resolve this?
", "rawMarkdown": "Were you able to resolve this? " }, { "id": 568730, "postDate": "2019-07-05T10:46:28.977Z", "content": "Not really. I tried reduce image resolution, and some other weird things, which didn't work. The funny thing is that when I changed only the line where results are assigned to random numbers (from 0 to 4) from numpy.random.randint the submission was fine (rest of the code was the same). We have also some other kernels where I cannot find huge differences and submission works. I just want to know what I am doing wrong in this specific kernel.
", "rawMarkdown": "Not really. I tried reduce image resolution, and some other weird things, which didn't work. The funny thing is that when I changed only the line where results are assigned to random numbers (from 0 to 4) from numpy.random.randint the submission was fine (rest of the code was the same). We have also some other kernels where I cannot find huge differences and submission works. I just want to know what I am doing wrong in this specific kernel." }, { "id": 569690, "postDate": "2019-07-07T06:58:54.917Z", "content": "This “running for 2 hours” can be your “editing kernel” not your “submitting kernel”. Note that when you make a commit, you create a separated kernel in addition to the editing version. The “editing kernel” can still run for 9 hours.
\n\nThey use the same name, so it may cause confusion.
", "rawMarkdown": "This “running for 2 hours” can be your “editing kernel” not your “submitting kernel”. Note that when you make a commit, you create a separated kernel in addition to the editing version. The “editing kernel” can still run for 9 hours.\n\nThey use the same name, so it may cause confusion.", "votes": 1 } ] } ], "comments": [ { "id": 569982, "author_name": "lippyt", "author_url": "", "post_date": "2019-07-07T16:26:52.983000", "content": "@solomonk I am encountering a similar problem. I have TTA implemented and my notebook outputs the submission CSV file after committing. However, the status eventually displays \"Submission CSV Not Found\".
\n\nI've tweaked the batch size in predict_generator and I can't get my kernel with TTA to work.
", "votes": 1, "replies": [ { "id": 570053, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-07T18:03:21.807000", "content": "For now, and since there was no response from the admins, I am not using TTA.
", "votes": 1, "replies": [] } ] }, { "id": 569939, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-07T15:16:15.833000", "content": "Once I removed the TTA I was able to submit. I can not fathom why this may impose a problem.
", "votes": 2, "replies": [ { "id": 570381, "author_name": "Federico Raimondi Cominesi", "author_url": "", "post_date": "2019-07-08T07:56:45.787000", "content": "Did you also reduce batch size or image size? Because I have the same CSV not found and I trying everything with no luck! I have the lazy TTA from Abhishek's Kernel but also without TTA isn't working.\nI m truly considering not investing more time on this competition...
", "votes": 1, "replies": [] }, { "id": 570615, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-08T14:43:54.740000", "content": "I can privately share a notebook that works without TTA.
", "votes": 0, "replies": [] }, { "id": 570636, "author_name": "Federico Raimondi Cominesi", "author_url": "", "post_date": "2019-07-08T15:08:09.803000", "content": "Pretty sure we can not share private code during the competition :smile:\nBut I would love to see a kernel with that (even a low score).
", "votes": 1, "replies": [] } ] }, { "id": 569297, "author_name": "David Edwards", "author_url": "", "post_date": "2019-07-06T11:38:51.690000", "content": "I am getting the same error too. My kernel is an inference version. Only takes a few minutes to run. I started two days ago for me. It started happening when I went to larger images, but reducing image size does not fix the issue.\nFixed it. The images had to be the same size I changed to 299 from 150, but not it has to be the original size to work, not 100% why yet.
", "votes": 2, "replies": [] }, { "id": 568506, "author_name": "Prashant Kikani", "author_url": "", "post_date": "2019-07-05T03:12:02.667000", "content": "I've faced same issue. I think its memory overflow. You can avoid it by reducing batch size (keeping same resolution) or reducing resolution (keeping same batch size).
", "votes": 2, "replies": [ { "id": 721014, "author_name": "", "author_url": "", "post_date": "2020-01-17T00:05:33.493000", "content": "I will try this method. I encountered the same issue in DFDC. There are 400 test videos and 4000 public test videos. Maybe memory is overflow.
", "votes": 0, "replies": [] } ] }, { "id": 567842, "author_name": "vbookshelf", "author_url": "", "post_date": "2019-07-04T04:37:43.840000", "content": "I'm getting this error when using larger image sizes and when using more training images. Therefore, I suspect that this is actually a memory related error. It could be that the system is running out of RAM during prediction. Reducing the image size may help.
", "votes": 2, "replies": [] }, { "id": 635261, "author_name": "Tony D", "author_url": "", "post_date": "2019-09-27T10:18:44.790000", "content": "I have the same issue with other dataset . I attached the noteboook
\n\nhttps://www.kaggle.com/questions-and-answers/110395
", "votes": 0, "replies": [] }, { "id": 615317, "author_name": "Rajni Jain", "author_url": "", "post_date": "2019-09-01T18:59:50.683000", "content": "I am also facing the same issue.If anyone has find the solution , Please share.
", "votes": 0, "replies": [] }, { "id": 610069, "author_name": "Vitor Alcântara Batista", "author_url": "", "post_date": "2019-08-28T11:34:28.180000", "content": "I'm having the same problem. My kernel runs in 2 minutes with GPU and 10 with CPU. The commit goes fine, producing the CSV in output. My code only loads the weights from a pre-trained model and predict test images. During execution, it consumes not more than 5GB of RAM.
", "votes": 0, "replies": [] }, { "id": 574831, "author_name": "Peter Nemeth", "author_url": "", "post_date": "2019-07-14T14:42:51.193000", "content": "Hi,
\n\nI got the same problem. The solution was for me to remove the file operations from the kernel.\nI used to save the preprocessed images first, before prediction.\nWhen I predicted the images on the fly, then I managed to submit the kernel.
\n\nSomething was wrong with saving the images to /tmp/
", "votes": 0, "replies": [] }, { "id": 574190, "author_name": "Peter Nemeth", "author_url": "", "post_date": "2019-07-13T11:02:01.473000", "content": "I had the same problem. \nI managed to solve it by removing all the file operations.
\n\nI used to save preprocessed images. Now I preprocess and predict everything on the fly, without saving them first to the disk. It helped...
", "votes": 0, "replies": [] }, { "id": 572077, "author_name": "Mirodil", "author_url": "", "post_date": "2019-07-10T12:19:44.987000", "content": "I have the same issue. Is anyone solved it?
", "votes": 0, "replies": [ { "id": 572082, "author_name": "Federico Raimondi Cominesi", "author_url": "", "post_date": "2019-07-10T12:39:40.953000", "content": "Yes, it should be a memory issue. In my case, it was either the image size or batch size during the training phase. TTA was not the cause.\nTry to reduce both and hopefully, it will work xD
", "votes": 0, "replies": [] } ] }, { "id": 569769, "author_name": "saran", "author_url": "", "post_date": "2019-07-07T09:35:56.177000", "content": "I too facing this problem...Please someone help me
", "votes": 0, "replies": [] }, { "id": 567743, "author_name": "Chanhu", "author_url": "", "post_date": "2019-07-04T00:40:08.407000", "content": "I have the same problem
", "votes": 0, "replies": [] }, { "id": 567504, "author_name": "omar", "author_url": "", "post_date": "2019-07-03T15:52:00.720000", "content": "I have the same problem, any ideas about how to solve it?
", "votes": 0, "replies": [ { "id": 567553, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-03T17:16:49.260000", "content": "I am trying to, once I have a solution I shall let you know.
", "votes": 0, "replies": [] } ] }, { "id": 567489, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-03T15:34:34.723000", "content": "Admins, kindly let me know the exact error or I shall have to quit the competition. No kidding.
", "votes": 0, "replies": [] }, { "id": 567343, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-03T11:16:37.720000", "content": "How can it be running for two hours, if the report stated that the submission file is not found ...?! \n
The same for me. I've tried different solutions written in other discussions but no results. I think that there should be some more specific info about error (lack of resources or anything else). It is a little frustrating to change code without knowing what went wrong.
", "votes": 2, "replies": [] }, { "id": 567490, "author_name": "QuantScientist", "author_url": "", "post_date": "2019-07-03T15:35:01.343000", "content": "Were you able to resolve this?
", "votes": 0, "replies": [] }, { "id": 568730, "author_name": "rafzy", "author_url": "", "post_date": "2019-07-05T10:46:28.977000", "content": "Not really. I tried reduce image resolution, and some other weird things, which didn't work. The funny thing is that when I changed only the line where results are assigned to random numbers (from 0 to 4) from numpy.random.randint the submission was fine (rest of the code was the same). We have also some other kernels where I cannot find huge differences and submission works. I just want to know what I am doing wrong in this specific kernel.
", "votes": 0, "replies": [] }, { "id": 569690, "author_name": "Neuron Engineer", "author_url": "", "post_date": "2019-07-07T06:58:54.917000", "content": "This “running for 2 hours” can be your “editing kernel” not your “submitting kernel”. Note that when you make a commit, you create a separated kernel in addition to the editing version. The “editing kernel” can still run for 9 hours.
\n\nThey use the same name, so it may cause confusion.
", "votes": 1, "replies": [] } ] } ], "raw_markdown_by_id": { "567317": "Dear All,\nMy kernel build successfully and the respective CSV is also created with the requested name:\n\n\nHowever time after time following submission, I immediately get this error:\n\n\nAny ideas as to what may be wrong? \nThanks. \n ", "569982": "@solomonk I am encountering a similar problem. I have TTA implemented and my notebook outputs the submission CSV file after committing. However, the status eventually displays \"Submission CSV Not Found\".\n\nI've tweaked the batch size in predict_generator and I can't get my kernel with TTA to work.", "569939": "Once I removed the TTA I was able to submit. I can not fathom why this may impose a problem. ", "569297": "I am getting the same error too. My kernel is an inference version. Only takes a few minutes to run. I started two days ago for me. It started happening when I went to larger images, but reducing image size does not fix the issue.\nFixed it. The images had to be the same size I changed to 299 from 150, but not it has to be the original size to work, not 100% why yet.", "568506": "I've faced same issue. I think its memory overflow. You can avoid it by reducing batch size (keeping same resolution) or reducing resolution (keeping same batch size).", "567842": "I'm getting this error when using larger image sizes and when using more training images. Therefore, I suspect that this is actually a memory related error. It could be that the system is running out of RAM during prediction. Reducing the image size may help.", "635261": "I have the same issue with other dataset . I attached the noteboook\n\nhttps://www.kaggle.com/questions-and-answers/110395\n\n", "615317": "I am also facing the same issue.If anyone has find the solution , Please share.", "610069": "I'm having the same problem. My kernel runs in 2 minutes with GPU and 10 with CPU. The commit goes fine, producing the CSV in output. My code only loads the weights from a pre-trained model and predict test images. During execution, it consumes not more than 5GB of RAM.", "574831": "Hi,\n\nI got the same problem. The solution was for me to remove the file operations from the kernel.\nI used to save the preprocessed images first, before prediction.\nWhen I predicted the images on the fly, then I managed to submit the kernel.\n\nSomething was wrong with saving the images to /tmp/", "574190": "I had the same problem. \nI managed to solve it by removing all the file operations.\n\nI used to save preprocessed images. Now I preprocess and predict everything on the fly, without saving them first to the disk. It helped...", "572077": "I have the same issue. Is anyone solved it?", "569769": "I too facing this problem...Please someone help me", "567743": "I have the same problem", "567504": "I have the same problem, any ideas about how to solve it?", "567489": "Admins, kindly let me know the exact error or I shall have to quit the competition. No kidding. ", "567343": "How can it be running for two hours, if the report stated that the submission file is not found ...?! \n\n" } }