I'm repeatedly getting submission csv not found error for the last 1 week even though the logs do not indicate any error and the submission csv is actually getting produced as per the logs. This is very frustrating, what am I supposed to do when there is no indication of error?
", "messages": [ { "id": "3348614", "postDate": "11/26/2025 04:29:42", "content": "I'm repeatedly getting submission csv not found error for the last 1 week even though the logs do not indicate any error and the submission csv is actually getting produced as per the logs. This is very frustrating, what am I supposed to do when there is no indication of error?
", "rawMarkdown": "I'm repeatedly getting submission csv not found error for the last 1 week even though the logs do not indicate any error and the submission csv is actually getting produced as per the logs. This is very frustrating, what am I supposed to do when there is no indication of error?", "votes": null }, { "id": "3349486", "postDate": "11/26/2025 16:43:46", "content": "submission.csv not found often means the notebook failed before the submission file was written while running the private test images. Remember that a leader board submission runs 800 images rather than the 1 image that we can see. Lots of issues can exist in code making predictions.
\nTODO:
\nItem 1 at kaggle is always this when this issue occurs.
\nIt should make the predictions against the 357 train images and will likely reveal errors in the cell - most often memory issues / batch size/ etc.
", "rawMarkdown": "submission.csv not found often means the notebook failed before the submission file was written while running the private test images. Remember that a leader board submission runs 800 images rather than the 1 image that we can see. Lots of issues can exist in code making predictions.\n\nTODO:\n\nItem 1 at kaggle is always this when this issue occurs.\n1. Open the notebook on kaggle.\n2. find where you read the test.csv file for the predictions.\n3. Replace test with train.csv.\n4. Run All\n\nIt should make the predictions against the 357 train images and will likely reveal errors in the cell - most often memory issues / batch size/ etc.", "votes": null }, { "id": "3351493", "postDate": "11/28/2025 13:03:51", "content": "The \"internal section\" mentioned in the paper refers to a 60cm x 30cm area within the original 70cm frame. By cropping out approximately 10cm of the width (likely 5cm from each side), the physical aspect ratio becomes 2:1 (60:30), which perfectly aligns with the digital resolution of 2,000 x 1,000 pixels.
", "rawMarkdown": "The \"internal section\" mentioned in the paper refers to a 60cm x 30cm area within the original 70cm frame. By cropping out approximately 10cm of the width (likely 5cm from each side), the physical aspect ratio becomes 2:1 (60:30), which perfectly aligns with the digital resolution of 2,000 x 1,000 pixels.", "votes": null } ], "comments": [ { "id": 3349486, "author_name": "pcjimmmy", "author_url": "", "post_date": "11/26/2025 16:43:46", "content": "submission.csv not found often means the notebook failed before the submission file was written while running the private test images. Remember that a leader board submission runs 800 images rather than the 1 image that we can see. Lots of issues can exist in code making predictions.
\nTODO:
\nItem 1 at kaggle is always this when this issue occurs.
\nIt should make the predictions against the 357 train images and will likely reveal errors in the cell - most often memory issues / batch size/ etc.
", "votes": null, "replies": [ { "id": 3351493, "author_name": "rustambazarbayev", "author_url": "", "post_date": "11/28/2025 13:03:51", "content": "The \"internal section\" mentioned in the paper refers to a 60cm x 30cm area within the original 70cm frame. By cropping out approximately 10cm of the width (likely 5cm from each side), the physical aspect ratio becomes 2:1 (60:30), which perfectly aligns with the digital resolution of 2,000 x 1,000 pixels.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "3348614": "I'm repeatedly getting submission csv not found error for the last 1 week even though the logs do not indicate any error and the submission csv is actually getting produced as per the logs. This is very frustrating, what am I supposed to do when there is no indication of error?", "3349486": "submission.csv not found often means the notebook failed before the submission file was written while running the private test images. Remember that a leader board submission runs 800 images rather than the 1 image that we can see. Lots of issues can exist in code making predictions.\n\nTODO:\n\nItem 1 at kaggle is always this when this issue occurs.\n1. Open the notebook on kaggle.\n2. find where you read the test.csv file for the predictions.\n3. Replace test with train.csv.\n4. Run All\n\nIt should make the predictions against the 357 train images and will likely reveal errors in the cell - most often memory issues / batch size/ etc.", "3351493": "The \"internal section\" mentioned in the paper refers to a 60cm x 30cm area within the original 70cm frame. By cropping out approximately 10cm of the width (likely 5cm from each side), the physical aspect ratio becomes 2:1 (60:30), which perfectly aligns with the digital resolution of 2,000 x 1,000 pixels." }, "source": "meta" }