![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1914802%2Fced0612b050643e2f904f5270a4a3651%2FCapture.PNG?generation=1579168909267833&alt=media\")
I got ‘Kaggle Error’ when trying to make a submission. Does anyone have an idea what it means? \nThanks!
", "messages": [ { "id": "718704", "postDate": "01/14/2020 17:13:12", "content": "I got ‘Kaggle Error’ when trying to make a submission. Does anyone have an idea what it means? \nThanks!
", "rawMarkdown": "I got ‘Kaggle Error’ when trying to make a submission. Does anyone have an idea what it means? \nThanks!", "votes": null }, { "id": "718753", "postDate": "01/14/2020 18:01:50", "content": "this could mean many things: \n- bug in your code(some of it didn't run as expected)\n- memory error(OOM)\n- shape mismatch of submission.csv when commiting
The most common problem is out of memory. You can debug it by committing your notebook using the train parquet files.
", "rawMarkdown": "The most common problem is out of memory. You can debug it by committing your notebook using the **train** parquet files.", "votes": null }, { "id": "718758", "postDate": "01/14/2020 18:11:43", "content": "@aleksandradeis The issue is majorly caused when the system gets out of memory.\nYou can probably try\n1. deleting large dataframes and recovering memory after that using GC.collect\n2. Try to load 1 parquet file process it and discard it then move to next file.
\n\nThere are lot of post mentioning about the same.
\n\nIf still you face the issue than share your notebook we will help you out.
\n\nGood luck
", "rawMarkdown": "aleksandradeis The issue is majorly caused when the system gets out of memory.\nYou can probably try\n1. deleting large dataframes and recovering memory after that using GC.collect\n2. Try to load 1 parquet file process it and discard it then move to next file.\n\nThere are lot of post mentioning about the same.\n\nIf still you face the issue than share your notebook we will help you out.\n\nGood luck", "votes": null }, { "id": "718761", "postDate": "01/14/2020 18:18:31", "content": "Thank you so much for your answer! I will try to run commit for train dataset
", "rawMarkdown": "Thank you so much for your answer! I will try to run commit for train dataset", "votes": null }, { "id": "718762", "postDate": "01/14/2020 18:20:23", "content": "Thank you so much! Deleting data frames with go collect is definitely worth trying!
\n\nI was just so confused by the error. Thought that Kaggle error might mean something specific.
", "rawMarkdown": "Thank you so much! Deleting data frames with go collect is definitely worth trying! \n\nI was just so confused by the error. Thought that Kaggle error might mean something specific.", "votes": null }, { "id": "718763", "postDate": "01/14/2020 18:21:34", "content": "Thank you! I’ll try to address these
", "rawMarkdown": "Thank you! I’ll try to address these", "votes": null }, { "id": "718767", "postDate": "01/14/2020 18:32:25", "content": "I think that means internal kaggle error. In the same case I successfully resubmitted kernel. \nIt seems to me In case out of memory I got \"Notebook Exceeded Allowed Compute\".
", "rawMarkdown": "I think that means internal kaggle error. In the same case I successfully resubmitted kernel. \nIt seems to me In case out of memory I got \"Notebook Exceeded Allowed Compute\".", "votes": null }, { "id": "719082", "postDate": "01/15/2020 05:42:13", "content": "Thanks! I resubmitted and got the other error 😄 It seems that Kaggle error is really some Kaggle crush which might correspond to anything
I am submitting the .csv file but it is failing and even resubmit is failing too. Second time also the submission is running for an hour but the score is ERROR and NO SCORE .
\n\nPlease can you guide me what is the problem with submission or kernel. Please review and guide, if any one could help.
\n\nhttps://www.kaggle.com/bhaveshthaker/bengali-ai-handwritten-grapheme-classification
\n\n
Read this nice guide : https://www.kaggle.com/c/bengaliai-cv19/discussion/126054
\n\nIn short: split your kernel on two new ones: train kernel and test kernel. Load models\\weights from train kernel to test kernel. Process test dataset in batch mode (try bs=128), donn't try load all test dataset in memory. To verify proper work test kernel commit it using train dataset instead test dataset.
", "rawMarkdown": "Read this nice guide : https://www.kaggle.com/c/bengaliai-cv19/discussion/126054\n\nIn short: split your kernel on two new ones: train kernel and test kernel. Load models\\weights from train kernel to test kernel. Process test dataset in batch mode (try bs=128), donn't try load all test dataset in memory. To verify proper work test kernel commit it using train dataset instead test dataset.", "votes": null } ], "comments": [ { "id": 718753, "author_name": "bibek777", "author_url": "", "post_date": "01/14/2020 18:01:50", "content": "this could mean many things: \n- bug in your code(some of it didn't run as expected)\n- memory error(OOM)\n- shape mismatch of submission.csv when commiting
Thank you! I’ll try to address these
", "votes": null, "replies": [] } ] }, { "id": 718754, "author_name": "pestipeti", "author_url": "", "post_date": "01/14/2020 18:02:02", "content": "The most common problem is out of memory. You can debug it by committing your notebook using the train parquet files.
", "votes": null, "replies": [] }, { "id": 718758, "author_name": "amit9484", "author_url": "", "post_date": "01/14/2020 18:11:43", "content": "@aleksandradeis The issue is majorly caused when the system gets out of memory.\nYou can probably try\n1. deleting large dataframes and recovering memory after that using GC.collect\n2. Try to load 1 parquet file process it and discard it then move to next file.
\n\nThere are lot of post mentioning about the same.
\n\nIf still you face the issue than share your notebook we will help you out.
\n\nGood luck
", "votes": null, "replies": [ { "id": 718762, "author_name": "aleksandradeis", "author_url": "", "post_date": "01/14/2020 18:20:23", "content": "Thank you so much! Deleting data frames with go collect is definitely worth trying!
\n\nI was just so confused by the error. Thought that Kaggle error might mean something specific.
", "votes": null, "replies": [] } ] }, { "id": 718761, "author_name": "aleksandradeis", "author_url": "", "post_date": "01/14/2020 18:18:31", "content": "Thank you so much for your answer! I will try to run commit for train dataset
", "votes": null, "replies": [] }, { "id": 718767, "author_name": "andreyzotov", "author_url": "", "post_date": "01/14/2020 18:32:25", "content": "I think that means internal kaggle error. In the same case I successfully resubmitted kernel. \nIt seems to me In case out of memory I got \"Notebook Exceeded Allowed Compute\".
", "votes": null, "replies": [ { "id": 719082, "author_name": "aleksandradeis", "author_url": "", "post_date": "01/15/2020 05:42:13", "content": "Thanks! I resubmitted and got the other error 😄 It seems that Kaggle error is really some Kaggle crush which might correspond to anything
I am submitting the .csv file but it is failing and even resubmit is failing too. Second time also the submission is running for an hour but the score is ERROR and NO SCORE .
\n\nPlease can you guide me what is the problem with submission or kernel. Please review and guide, if any one could help.
\n\nhttps://www.kaggle.com/bhaveshthaker/bengali-ai-handwritten-grapheme-classification
\n\n
Read this nice guide : https://www.kaggle.com/c/bengaliai-cv19/discussion/126054
\n\nIn short: split your kernel on two new ones: train kernel and test kernel. Load models\\weights from train kernel to test kernel. Process test dataset in batch mode (try bs=128), donn't try load all test dataset in memory. To verify proper work test kernel commit it using train dataset instead test dataset.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "718704": "I got ‘Kaggle Error’ when trying to make a submission. Does anyone have an idea what it means? \nThanks!", "718753": "this could mean many things: \n- bug in your code(some of it didn't run as expected)\n- memory error(OOM)\n- shape mismatch of `submission.csv` when commiting", "718754": "The most common problem is out of memory. You can debug it by committing your notebook using the **train** parquet files.", "718758": "aleksandradeis The issue is majorly caused when the system gets out of memory.\nYou can probably try\n1. deleting large dataframes and recovering memory after that using GC.collect\n2. Try to load 1 parquet file process it and discard it then move to next file.\n\nThere are lot of post mentioning about the same.\n\nIf still you face the issue than share your notebook we will help you out.\n\nGood luck", "718761": "Thank you so much for your answer! I will try to run commit for train dataset", "718762": "Thank you so much! Deleting data frames with go collect is definitely worth trying! \n\nI was just so confused by the error. Thought that Kaggle error might mean something specific.", "718763": "Thank you! I’ll try to address these", "718767": "I think that means internal kaggle error. In the same case I successfully resubmitted kernel. \nIt seems to me In case out of memory I got \"Notebook Exceeded Allowed Compute\".", "719082": "Thanks! I resubmitted and got the other error 😄 It seems that `Kaggle error` is really some Kaggle crush which might correspond to anything", "720300": "I am submitting the .csv file but it is failing and even resubmit is failing too. Second time also the submission is running for an hour but the score is ERROR and NO SCORE .\n\nPlease can you guide me what is the problem with submission or kernel. Please review and guide, if any one could help.\n\nhttps://www.kaggle.com/bhaveshthaker/bengali-ai-handwritten-grapheme-classification\n\n\n\n", "720391": "Read this nice guide : https://www.kaggle.com/c/bengaliai-cv19/discussion/126054\n\nIn short: split your kernel on two new ones: train kernel and test kernel. Load models\\weights from train kernel to test kernel. Process test dataset in batch mode (try bs=128), donn't try load all test dataset in memory. To verify proper work test kernel commit it using train dataset instead test dataset." }, "source": "meta" }