Hello! My team has been having a lot of…\"fun\" trying to debug our submission notebook over the past week. Well over half of our total submissions so far have been burned up just trying to figure out what is triggering our notebook to die only on the hidden test set. At first the errors all said \"Notebook Threw Exception,\" but now we are facing the far more elusive \"Submission CSV Not Found.\" It's frustrating that we can't tell what this means; it's not an Out Of Memory or Notebook Raised Exception error, but that doesn't even mean we can entirely rule those possibilities out.
\nI'm sure that we're not the only ones who have struggled--or will continue to struggle--to debug this particular error message. So how about we make a list of observations on this total black box of an error as a resource to help each other get bast this awful barrier?
\nI've come across this issue in past contests before, but I'm still not sure what triggers it. I wonder if a kernel crash could cause it, since a kernel crash might be recorded as neither an exception nor an Out Of Memory error. The notebook certainly does terminate before it gets to the part where it outputs the submission.csv. What does seem certain is that the kernel stops for one reason or another, but the error does not get properly detected and displayed to the user, with the scorer instead attempting to find the Submission CSV.
", "messages": [ { "id": 2499428, "postDate": "2023-10-26T03:21:56.707Z", "content": "Hello! My team has been having a lot of…\"fun\" trying to debug our submission notebook over the past week. Well over half of our total submissions so far have been burned up just trying to figure out what is triggering our notebook to die only on the hidden test set. At first the errors all said \"Notebook Threw Exception,\" but now we are facing the far more elusive \"Submission CSV Not Found.\" It's frustrating that we can't tell what this means; it's not an Out Of Memory or Notebook Raised Exception error, but that doesn't even mean we can entirely rule those possibilities out.
\nI'm sure that we're not the only ones who have struggled--or will continue to struggle--to debug this particular error message. So how about we make a list of observations on this total black box of an error as a resource to help each other get bast this awful barrier?
\nI've come across this issue in past contests before, but I'm still not sure what triggers it. I wonder if a kernel crash could cause it, since a kernel crash might be recorded as neither an exception nor an Out Of Memory error. The notebook certainly does terminate before it gets to the part where it outputs the submission.csv. What does seem certain is that the kernel stops for one reason or another, but the error does not get properly detected and displayed to the user, with the scorer instead attempting to find the Submission CSV.
", "rawMarkdown": "Hello! My team has been having a lot of...\"fun\" trying to debug our submission notebook over the past week. Well over half of our total submissions so far have been burned up just trying to figure out what is triggering our notebook to die only on the hidden test set. At first the errors all said \"Notebook Threw Exception,\" but now we are facing the far more elusive \"Submission CSV Not Found.\" It's frustrating that we can't tell what this means; it's not an Out Of Memory or Notebook Raised Exception error, but that doesn't even mean we can entirely rule those possibilities out.\n\nI'm sure that we're not the only ones who have struggled--or will continue to struggle--to debug this particular error message. So how about we make a list of observations on this total black box of an error as a resource to help each other get bast this awful barrier?\n\nI've come across this issue in past contests before, but I'm still not sure what triggers it. I wonder if a kernel crash could cause it, since a kernel crash might be recorded as neither an exception nor an Out Of Memory error. The notebook certainly does terminate before it gets to the part where it outputs the submission.csv. What does seem certain is that the kernel stops for one reason or another, but the error does not get properly detected and displayed to the user, with the scorer instead attempting to find the Submission CSV.", "votes": 5 }, { "id": 2499598, "postDate": "2023-10-26T06:17:27.457Z", "content": "I just encountered it for the first time right now. I believe the issue originated when I was preprocessing the test images and saving them at '/kaggle/temp/images.' I then changed this to '/kaggle/working/images,' and now the submission CSV file cannot be found.
\nSo you can try the following:
\nsave the preprocessed images in a temp directory outside /kaggle/working, and in the last cell do this:
os.listdir():\n sub.to_csv(, index = )\nYeah, same here. Files/folders other than \"submission.csv\" in \"/kaggle/working\" caused me this error as well. Removing them solved it.
", "rawMarkdown": "Yeah, same here. Files/folders other than \"submission.csv\" in \"/kaggle/working\" caused me this error as well. Removing them solved it.", "votes": 1 } ] } ], "comments": [ { "id": 2499598, "author_name": "pjmathematician", "author_url": "", "post_date": "2023-10-26T06:17:27.457000", "content": "I just encountered it for the first time right now. I believe the issue originated when I was preprocessing the test images and saving them at '/kaggle/temp/images.' I then changed this to '/kaggle/working/images,' and now the submission CSV file cannot be found.
\nSo you can try the following:
\nsave the preprocessed images in a temp directory outside /kaggle/working, and in the last cell do this:
os.listdir():\n sub.to_csv(, index = )\nYeah, same here. Files/folders other than \"submission.csv\" in \"/kaggle/working\" caused me this error as well. Removing them solved it.
", "votes": 1, "replies": [] } ] } ], "raw_markdown_by_id": { "2499428": "Hello! My team has been having a lot of...\"fun\" trying to debug our submission notebook over the past week. Well over half of our total submissions so far have been burned up just trying to figure out what is triggering our notebook to die only on the hidden test set. At first the errors all said \"Notebook Threw Exception,\" but now we are facing the far more elusive \"Submission CSV Not Found.\" It's frustrating that we can't tell what this means; it's not an Out Of Memory or Notebook Raised Exception error, but that doesn't even mean we can entirely rule those possibilities out.\n\nI'm sure that we're not the only ones who have struggled--or will continue to struggle--to debug this particular error message. So how about we make a list of observations on this total black box of an error as a resource to help each other get bast this awful barrier?\n\nI've come across this issue in past contests before, but I'm still not sure what triggers it. I wonder if a kernel crash could cause it, since a kernel crash might be recorded as neither an exception nor an Out Of Memory error. The notebook certainly does terminate before it gets to the part where it outputs the submission.csv. What does seem certain is that the kernel stops for one reason or another, but the error does not get properly detected and displayed to the user, with the scorer instead attempting to find the Submission CSV.", "2499598": "I just encountered it for the first time right now. I believe the issue originated when I was preprocessing the test images and saving them at '/kaggle/temp/images.' I then changed this to '/kaggle/working/images,' and now the submission CSV file cannot be found. \n\nSo you can try the following:\nsave the preprocessed images in a temp directory outside /kaggle/working, and in the last cell do this:\n```python\nwhile \"submission.csv\" not in os.listdir(\"/kaggle/working\"):\n sub.to_csv('/kaggle/working/submission.csv', index = False)\n```\n" } }