After the first submission failed, I used the training set as the test set to check the program and found the problem, but the second submission also showed that no CSV file could be found. Has anyone encountered this problem
", "messages": [ { "id": "1272117", "postDate": "04/13/2021 08:32:34", "content": "After the first submission failed, I used the training set as the test set to check the program and found the problem, but the second submission also showed that no CSV file could be found. Has anyone encountered this problem
", "rawMarkdown": "After the first submission failed, I used the training set as the test set to check the program and found the problem, but the second submission also showed that no CSV file could be found. Has anyone encountered this problem", "votes": null }, { "id": "1272213", "postDate": "04/13/2021 10:06:44", "content": "I have encountered the same problem.. Most probably you have written many files along with the submission.csv in the output.
", "rawMarkdown": "I have encountered the same problem.. Most probably you have written many files along with the *submission.csv* in the output.", "votes": null }, { "id": "1272222", "postDate": "04/13/2021 10:18:05", "content": "@onucsecu I only have one CSV output file
", "rawMarkdown": "onucsecu I only have one CSV output file", "votes": null }, { "id": "1272233", "postDate": "04/13/2021 10:28:06", "content": "@dlidli Then I have no idea. I am also looking for the solution. First i had this problem later I just let have one output which is submission.csv. now getting Submission Scoring Error.. but everything is seemed fine when i run it. ๐ฃ
", "rawMarkdown": "dlidli Then I have no idea. I am also looking for the solution. First i had this problem later I just let have one output which is submission.csv. now getting Submission Scoring Error.. but everything is seemed fine when i run it. ๐ฃ", "votes": null }, { "id": "1272236", "postDate": "04/13/2021 10:33:02", "content": "@onucsecu Me too. I input the training set as the test set into the network and can also produce CSV files. However, it shows that the CSV files cannot be found after submission
", "rawMarkdown": "onucsecu Me too. I input the training set as the test set into the network and can also produce CSV files. However, it shows that the CSV files cannot be found after submission", "votes": null }, { "id": "1272242", "postDate": "04/13/2021 10:39:35", "content": "its very weird.. Kaggle should give us more specific error log :(
", "rawMarkdown": "its very weird.. Kaggle should give us more specific error log :(", "votes": null }, { "id": "1272247", "postDate": "04/13/2021 10:47:26", "content": "You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large
", "rawMarkdown": "You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large", "votes": null }, { "id": "1272250", "postDate": "04/13/2021 10:47:49", "content": "@onucsecu You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large
", "rawMarkdown": "onucsecu You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large", "votes": null }, { "id": "1272381", "postDate": "04/13/2021 12:36:33", "content": "This means that your notebook produces an error while running on the full test set. In this case it does not able to generate csv file in the end because stops earlier and that is why you are getting this error message. You can try to reproduce it by manually increasing number of test images while committing the notebook. Most probably you have an issue related to out of memory on RAM or on GPU, which may be not detected while you are processing only three images.
", "rawMarkdown": "This means that your notebook produces an error while running on the full test set. In this case it does not able to generate csv file in the end because stops earlier and that is why you are getting this error message. You can try to reproduce it by manually increasing number of test images while committing the notebook. Most probably you have an issue related to out of memory on RAM or on GPU, which may be not detected while you are processing only three images.", "votes": null }, { "id": "1272435", "postDate": "04/13/2021 13:24:24", "content": "@opanichev Thank you for your reply. I'll have a try๐ช
", "rawMarkdown": "opanichev Thank you for your reply. I'll have a try๐ช", "votes": null }, { "id": "1279630", "postDate": "04/21/2021 05:44:37", "content": "@opanichev Finally, I found that every time I would read the test CSV file and put the test results on the test CSV file. After submission, the test data set was too long to match the length of the original 3. Thanks for your answer.
", "rawMarkdown": "opanichev Finally, I found that every time I would read the test CSV file and put the test results on the test CSV file. After submission, the test data set was too long to match the length of the original 3. Thanks for your answer.", "votes": null }, { "id": "1282627", "postDate": "04/24/2021 06:14:21", "content": "Finally i solved the problem what i did was that i opened the notebook and pressed run all and in one of my cells there was an error i misplaced argmax with argmaz and all the other code including where i exported submission were getting errors so it was not being exported kaggle surely should provide us better logs . Even if you dont get an error which actually i also did not got(in another notebook)because before argmax there was an if statement and none of values in the test set of 3 images met the condition(but in 5000 images it met) so just recheck your code
", "rawMarkdown": "Finally i solved the problem what i did was that i opened the notebook and pressed run all and in one of my cells there was an error i misplaced argmax with argmaz and all the other code including where i exported submission were getting errors so it was not being exported kaggle surely should provide us better logs . Even if you dont get an error which actually i also did not got(in another notebook)because before argmax there was an if statement and none of values in the test set of 3 images met the condition(but in 5000 images it met) so just recheck your code", "votes": null }, { "id": "1315909", "postDate": "05/20/2021 07:22:49", "content": "@jl18pg052 this information may help you.
", "rawMarkdown": "jl18pg052 this information may help you.", "votes": null }, { "id": "1315935", "postDate": "05/20/2021 07:45:57", "content": "@onucsecu Thank you for landing me to the right place manโฆ๐๐
", "rawMarkdown": "onucsecu Thank you for landing me to the right place man...๐๐", "votes": null }, { "id": "1315950", "postDate": "05/20/2021 07:52:47", "content": "@opanichev the point that you mentioned that it may be running out of memory that is GPU, it is a possibility, cause when i was training it on Kaggle, it was going out of memory on GPU, cause I was using larger image size, as I am training on efficientnetB7 model, which takes larger image size, but now I will try to train it on smaller image size, and then train it.
", "rawMarkdown": "opanichev the point that you mentioned that it may be running out of memory that is GPU, it is a possibility, cause when i was training it on Kaggle, it was going out of memory on GPU, cause I was using larger image size, as I am training on efficientnetB7 model, which takes larger image size, but now I will try to train it on smaller image size, and then train it.", "votes": null }, { "id": "1315972", "postDate": "05/20/2021 08:11:51", "content": "@jl18pg052 find the root cause of the problem you can take 5000 images from train data as test data and run your notebook again.
\ni suggest you to reduce batch size if there are any possibility of the out of GPU memory and use pretrained model.
Yeahโฆ I am trying that
", "rawMarkdown": "Yeah... I am trying that", "votes": null } ], "comments": [ { "id": 1272213, "author_name": "onucsecu", "author_url": "", "post_date": "04/13/2021 10:06:44", "content": "I have encountered the same problem.. Most probably you have written many files along with the submission.csv in the output.
", "votes": null, "replies": [] }, { "id": 1272222, "author_name": "dlidli", "author_url": "", "post_date": "04/13/2021 10:18:05", "content": "@onucsecu I only have one CSV output file
", "votes": null, "replies": [] }, { "id": 1272233, "author_name": "onucsecu", "author_url": "", "post_date": "04/13/2021 10:28:06", "content": "@dlidli Then I have no idea. I am also looking for the solution. First i had this problem later I just let have one output which is submission.csv. now getting Submission Scoring Error.. but everything is seemed fine when i run it. ๐ฃ
", "votes": null, "replies": [] }, { "id": 1272236, "author_name": "dlidli", "author_url": "", "post_date": "04/13/2021 10:33:02", "content": "@onucsecu Me too. I input the training set as the test set into the network and can also produce CSV files. However, it shows that the CSV files cannot be found after submission
", "votes": null, "replies": [ { "id": 1272242, "author_name": "onucsecu", "author_url": "", "post_date": "04/13/2021 10:39:35", "content": "its very weird.. Kaggle should give us more specific error log :(
", "votes": null, "replies": [] } ] }, { "id": 1272247, "author_name": "dlidli", "author_url": "", "post_date": "04/13/2021 10:47:26", "content": "You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large
", "votes": null, "replies": [] }, { "id": 1272250, "author_name": "dlidli", "author_url": "", "post_date": "04/13/2021 10:47:49", "content": "@onucsecu You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large
", "votes": null, "replies": [] }, { "id": 1272381, "author_name": "opanichev", "author_url": "", "post_date": "04/13/2021 12:36:33", "content": "This means that your notebook produces an error while running on the full test set. In this case it does not able to generate csv file in the end because stops earlier and that is why you are getting this error message. You can try to reproduce it by manually increasing number of test images while committing the notebook. Most probably you have an issue related to out of memory on RAM or on GPU, which may be not detected while you are processing only three images.
", "votes": null, "replies": [ { "id": 1315909, "author_name": "onucsecu", "author_url": "", "post_date": "05/20/2021 07:22:49", "content": "@jl18pg052 this information may help you.
", "votes": null, "replies": [] }, { "id": 1315935, "author_name": "jl18pg052", "author_url": "", "post_date": "05/20/2021 07:45:57", "content": "@onucsecu Thank you for landing me to the right place manโฆ๐๐
", "votes": null, "replies": [] }, { "id": 1315950, "author_name": "jl18pg052", "author_url": "", "post_date": "05/20/2021 07:52:47", "content": "@opanichev the point that you mentioned that it may be running out of memory that is GPU, it is a possibility, cause when i was training it on Kaggle, it was going out of memory on GPU, cause I was using larger image size, as I am training on efficientnetB7 model, which takes larger image size, but now I will try to train it on smaller image size, and then train it.
", "votes": null, "replies": [] }, { "id": 1315972, "author_name": "onucsecu", "author_url": "", "post_date": "05/20/2021 08:11:51", "content": "@jl18pg052 find the root cause of the problem you can take 5000 images from train data as test data and run your notebook again.
\ni suggest you to reduce batch size if there are any possibility of the out of GPU memory and use pretrained model.
Yeahโฆ I am trying that
", "votes": null, "replies": [] } ] }, { "id": 1272435, "author_name": "dlidli", "author_url": "", "post_date": "04/13/2021 13:24:24", "content": "@opanichev Thank you for your reply. I'll have a try๐ช
", "votes": null, "replies": [] }, { "id": 1279630, "author_name": "dlidli", "author_url": "", "post_date": "04/21/2021 05:44:37", "content": "@opanichev Finally, I found that every time I would read the test CSV file and put the test results on the test CSV file. After submission, the test data set was too long to match the length of the original 3. Thanks for your answer.
", "votes": null, "replies": [] }, { "id": 1282627, "author_name": "swaralipibose", "author_url": "", "post_date": "04/24/2021 06:14:21", "content": "Finally i solved the problem what i did was that i opened the notebook and pressed run all and in one of my cells there was an error i misplaced argmax with argmaz and all the other code including where i exported submission were getting errors so it was not being exported kaggle surely should provide us better logs . Even if you dont get an error which actually i also did not got(in another notebook)because before argmax there was an if statement and none of values in the test set of 3 images met the condition(but in 5000 images it met) so just recheck your code
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1272117": "After the first submission failed, I used the training set as the test set to check the program and found the problem, but the second submission also showed that no CSV file could be found. Has anyone encountered this problem", "1272213": "I have encountered the same problem.. Most probably you have written many files along with the *submission.csv* in the output.", "1272222": "onucsecu I only have one CSV output file", "1272233": "dlidli Then I have no idea. I am also looking for the solution. First i had this problem later I just let have one output which is submission.csv. now getting Submission Scoring Error.. but everything is seemed fine when i run it. ๐ฃ", "1272236": "onucsecu Me too. I input the training set as the test set into the network and can also produce CSV files. However, it shows that the CSV files cannot be found after submission", "1272242": "its very weird.. Kaggle should give us more specific error log :(", "1272247": "You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large", "1272250": "onucsecu You've tried to run the training set as a test set, but I've seen it sometimes because it hides the problem of the training set being too large", "1272381": "This means that your notebook produces an error while running on the full test set. In this case it does not able to generate csv file in the end because stops earlier and that is why you are getting this error message. You can try to reproduce it by manually increasing number of test images while committing the notebook. Most probably you have an issue related to out of memory on RAM or on GPU, which may be not detected while you are processing only three images.", "1272435": "opanichev Thank you for your reply. I'll have a try๐ช", "1279630": "opanichev Finally, I found that every time I would read the test CSV file and put the test results on the test CSV file. After submission, the test data set was too long to match the length of the original 3. Thanks for your answer.", "1282627": "Finally i solved the problem what i did was that i opened the notebook and pressed run all and in one of my cells there was an error i misplaced argmax with argmaz and all the other code including where i exported submission were getting errors so it was not being exported kaggle surely should provide us better logs . Even if you dont get an error which actually i also did not got(in another notebook)because before argmax there was an if statement and none of values in the test set of 3 images met the condition(but in 5000 images it met) so just recheck your code", "1315909": "jl18pg052 this information may help you.", "1315935": "onucsecu Thank you for landing me to the right place man...๐๐", "1315950": "opanichev the point that you mentioned that it may be running out of memory that is GPU, it is a possibility, cause when i was training it on Kaggle, it was going out of memory on GPU, cause I was using larger image size, as I am training on efficientnetB7 model, which takes larger image size, but now I will try to train it on smaller image size, and then train it.", "1315972": "jl18pg052 find the root cause of the problem you can take 5000 images from train data as test data and run your notebook again.\ni suggest you to reduce batch size if there are any possibility of the out of GPU memory and use pretrained model.", "1315979": "Yeah... I am trying that" }, "source": "meta" }