Hi guys, I'm running into a problem that I can save a new version of my notebook successfully. But when I use this notebook for submitting, Kaggle can not score my submission and it threw an error that \"Notebook Threw Exception\". I use CodeBert based and 10000 samples for my training process. Can anyone help me with this problem? Thanks a lot.
", "messages": [ { "id": "1836370", "postDate": "06/28/2022 15:14:58", "content": "Hi guys, I'm running into a problem that I can save a new version of my notebook successfully. But when I use this notebook for submitting, Kaggle can not score my submission and it threw an error that \"Notebook Threw Exception\". I use CodeBert based and 10000 samples for my training process. Can anyone help me with this problem? Thanks a lot.
", "rawMarkdown": "Hi guys, I'm running into a problem that I can save a new version of my notebook successfully. But when I use this notebook for submitting, Kaggle can not score my submission and it threw an error that \"Notebook Threw Exception\". I use CodeBert based and 10000 samples for my training process. Can anyone help me with this problem? Thanks a lot.", "votes": null }, { "id": "1836986", "postDate": "06/29/2022 08:49:43", "content": "I encountered this one once. In my case, my preprocessing code uses more RAM than the kaggle allowed so it crashed. Maybe you should check your implementation, make sure it can run before submitting.
", "rawMarkdown": "I encountered this one once. In my case, my preprocessing code uses more RAM than the kaggle allowed so it crashed. Maybe you should check your implementation, make sure it can run before submitting.", "votes": null }, { "id": "1836996", "postDate": "06/29/2022 08:54:55", "content": "If I used more RAM than allowed, I think Kaggle would throw an error called \"Notebook Exceeded Allowed Compute\" not the \"Notebook Threw Exception\" statement.
", "rawMarkdown": "If I used more RAM than allowed, I think Kaggle would throw an error called \"Notebook Exceeded Allowed Compute\" not the \"Notebook Threw Exception\" statement.", "votes": null }, { "id": "1837839", "postDate": "06/30/2022 01:52:47", "content": "In the notebook environment in your submission section, try feeding it 20-25,000 of the training samples instead - ideally a local validation set if you have one. That got me through one of my bugs - I was looking at only cell ID, which works fine with you only run the 4 test samples provided locally but it fails when you run enough notebooks through it.
\nIn addition, you can then take the output of that training-as-test run and verify the output makes sense by comparing it against ground truth.
", "rawMarkdown": "In the notebook environment in your submission section, try feeding it 20-25,000 of the training samples instead - ideally a local validation set if you have one. That got me through one of my bugs - I was looking at only cell ID, which works fine with you only run the 4 test samples provided locally but it fails when you run enough notebooks through it.\n\nIn addition, you can then take the output of that training-as-test run and verify the output makes sense by comparing it against ground truth.", "votes": null } ], "comments": [ { "id": 1836986, "author_name": "locbaop", "author_url": "", "post_date": "06/29/2022 08:49:43", "content": "I encountered this one once. In my case, my preprocessing code uses more RAM than the kaggle allowed so it crashed. Maybe you should check your implementation, make sure it can run before submitting.
", "votes": null, "replies": [ { "id": 1836996, "author_name": "nguynthhi", "author_url": "", "post_date": "06/29/2022 08:54:55", "content": "If I used more RAM than allowed, I think Kaggle would throw an error called \"Notebook Exceeded Allowed Compute\" not the \"Notebook Threw Exception\" statement.
", "votes": null, "replies": [] } ] }, { "id": 1837839, "author_name": "happycube", "author_url": "", "post_date": "06/30/2022 01:52:47", "content": "In the notebook environment in your submission section, try feeding it 20-25,000 of the training samples instead - ideally a local validation set if you have one. That got me through one of my bugs - I was looking at only cell ID, which works fine with you only run the 4 test samples provided locally but it fails when you run enough notebooks through it.
\nIn addition, you can then take the output of that training-as-test run and verify the output makes sense by comparing it against ground truth.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1836370": "Hi guys, I'm running into a problem that I can save a new version of my notebook successfully. But when I use this notebook for submitting, Kaggle can not score my submission and it threw an error that \"Notebook Threw Exception\". I use CodeBert based and 10000 samples for my training process. Can anyone help me with this problem? Thanks a lot.", "1836986": "I encountered this one once. In my case, my preprocessing code uses more RAM than the kaggle allowed so it crashed. Maybe you should check your implementation, make sure it can run before submitting.", "1836996": "If I used more RAM than allowed, I think Kaggle would throw an error called \"Notebook Exceeded Allowed Compute\" not the \"Notebook Threw Exception\" statement.", "1837839": "In the notebook environment in your submission section, try feeding it 20-25,000 of the training samples instead - ideally a local validation set if you have one. That got me through one of my bugs - I was looking at only cell ID, which works fine with you only run the 4 test samples provided locally but it fails when you run enough notebooks through it.\n\nIn addition, you can then take the output of that training-as-test run and verify the output makes sense by comparing it against ground truth." }, "source": "meta" }