![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4634683%2F3c4139c81bce45280be7f3b2ffface50%2F_20250527204944.png?generation=1748350200358551&alt=media\")
It's my first time participating and I keep getting the 'submission scoring error' Which I think is due to invalid file format of my output compared to sample_submission. However I mannually and through code compared both the documents and couldn't seem to find any issues, same columns, rows, no header errors, spelling mistakes, encoding issues (utf-8), same datatypes for every column, unnecessary spaces; just replaced with my prediction data. Same index, spelling of headers etc.
\nAfter multiple attempts I couldn't make it work. So after listening to someone's advice I just uploaded and submitted the sample_submission file with all zeros in the prediction columns. And that gave me a score. So I then took the sample_submission file and edited it to replace one of the zero columns (x_1) with both -1 and 1 (still float64 dtype) and then submitted again. And both of those failed. So I think something is wrong with placing numbers on the columns or something? Please help me how I can fix this scoring issue.
", "messages": [ { "id": "3206280", "postDate": "05/21/2025 05:43:10", "content": "It's my first time participating and I keep getting the 'submission scoring error' Which I think is due to invalid file format of my output compared to sample_submission. However I mannually and through code compared both the documents and couldn't seem to find any issues, same columns, rows, no header errors, spelling mistakes, encoding issues (utf-8), same datatypes for every column, unnecessary spaces; just replaced with my prediction data. Same index, spelling of headers etc.
\nAfter multiple attempts I couldn't make it work. So after listening to someone's advice I just uploaded and submitted the sample_submission file with all zeros in the prediction columns. And that gave me a score. So I then took the sample_submission file and edited it to replace one of the zero columns (x_1) with both -1 and 1 (still float64 dtype) and then submitted again. And both of those failed. So I think something is wrong with placing numbers on the columns or something? Please help me how I can fix this scoring issue.
", "rawMarkdown": "It's my first time participating and I keep getting the 'submission scoring error' Which I think is due to invalid file format of my output compared to sample_submission. However I mannually and through code compared both the documents and couldn't seem to find any issues, same columns, rows, no header errors, spelling mistakes, encoding issues (utf-8), same datatypes for every column, unnecessary spaces; just replaced with my prediction data. Same index, spelling of headers etc. \n\nAfter multiple attempts I couldn't make it work. So after listening to someone's advice I just uploaded and submitted the sample_submission file with all zeros in the prediction columns. And that gave me a score. So I then took the sample_submission file and edited it to replace one of the zero columns (x_1) with both -1 and 1 (still float64 dtype) and then submitted again. And both of those failed. So I think something is wrong with placing numbers on the columns or something? Please help me how I can fix this scoring issue.", "votes": null }, { "id": "3210501", "postDate": "05/27/2025 09:35:41", "content": "I'm experiencing a similar submission problem. I've verified that the sample_submission.csv goes through the scoring procedure, and my own submission CSV also works correctly with the official scoring function by comparing it against the sample_submission.csv. But the formal submission still failed. Since I noticed you have a successful ranking, I was hoping you might have overcome this issue and could offer some advice?
after prediction use sample_submission.csv to match the rows
", "rawMarkdown": "after prediction use sample_submission.csv to match the rows", "votes": null }, { "id": "3210613", "postDate": "05/27/2025 12:51:51", "content": "Thanks for the tip. I've tried it, but the same error persists.
if you use this \"Using to_csv(output_filename, index=False, mode='a', header=(i == 0)) to save submission rows can cause issues if the model crashes mid-way, resulting in an incomplete submission file. To avoid this, wrap the save logic in a try-except block so the error is skipped and at end use sample _submisison.csv to fine check where the skipped part will be replace and hence you get complete csv
", "rawMarkdown": "if you use this \"Using to_csv(output_filename, index=False, mode='a', header=(i == 0)) to save submission rows can cause issues if the model crashes mid-way, resulting in an incomplete submission file. To avoid this, wrap the save logic in a try-except block so the error is skipped and at end use sample _submisison.csv to fine check where the skipped part will be replace and hence you get complete csv", "votes": null }, { "id": "3210760", "postDate": "05/27/2025 16:49:22", "content": "I have a successful ranking because since nothing worked, i just submitted the sample_submission file as is, with 0s in all the columns. I still haven't been able to solve it. If you have, please let me know
", "rawMarkdown": "I have a successful ranking because since nothing worked, i just submitted the sample_submission file as is, with 0s in all the columns. I still haven't been able to solve it. If you have, please let me know", "votes": null }, { "id": "3210761", "postDate": "05/27/2025 16:51:40", "content": "Same, i also have the same dimensions of prediction dataset, doesn't seem to work
", "rawMarkdown": "Same, i also have the same dimensions of prediction dataset, doesn't seem to work", "votes": null }, { "id": "3210767", "postDate": "05/27/2025 16:56:32", "content": "I've only had one successful submission so far, and that was by skipping predictions for targets longer than 300 residues (R1126, R1136, R1138 – I used 'all zeros' for them), with no other modifications. If I try to include the predictions for these three, which my model can generate without issue, it still causes an error. It's quite annoying, as the final score is an overall average across all targets.
", "rawMarkdown": "I've only had one successful submission so far, and that was by skipping predictions for targets longer than 300 residues (R1126, R1136, R1138 – I used 'all zeros' for them), with no other modifications. If I try to include the predictions for these three, which my model can generate without issue, it still causes an error. It's quite annoying, as the final score is an overall average across all targets.", "votes": null }, { "id": "3210768", "postDate": "05/27/2025 16:56:59", "content": "Well mine says a scoring error, so i imagine there's something up with the formatting of the file rather than the model crashing mid way. Still not sure where its going wrong, the formatting looks all the same to me
", "rawMarkdown": "Well mine says a scoring error, so i imagine there's something up with the formatting of the file rather than the model crashing mid way. Still not sure where its going wrong, the formatting looks all the same to me", "votes": null }, { "id": "3210774", "postDate": "05/27/2025 16:59:45", "content": "That's honestly really weird and i'm not sure if its your fault tbh, let me try the same thing and see if i get a scoring error. This is my first kaggle competition and this has me in low energy tbh. Would be a shame if there's some issue with the scoring system itself if certain target ID's don't take predictions at all.
", "rawMarkdown": "That's honestly really weird and i'm not sure if its your fault tbh, let me try the same thing and see if i get a scoring error. This is my first kaggle competition and this has me in low energy tbh. Would be a shame if there's some issue with the scoring system itself if certain target ID's don't take predictions at all.", "votes": null }, { "id": "3210823", "postDate": "05/27/2025 17:48:06", "content": "you have a different testset , not the one available publicly, so if you script works well on public testset because the sequence length is not more than 720 , but in private testset it might be longer …
", "rawMarkdown": "you have a different testset , not the one available publicly, so if you script works well on public testset because the sequence length is not more than 720 , but in private testset it might be longer ...", "votes": null }, { "id": "3210841", "postDate": "05/27/2025 18:12:27", "content": "That makes sense, and yes, I got a successful submission by setting a length limit. But the weird thing is, if I just submit the plain sample_submission.csv (only containing the public testset, the notebook did not run any predictions), it still goes through the scoring process just fine.
Update: Tried it, it didn't work. Could not make the submission work no matter what. Extremely disappointing as my first kaggle competition. I got absolutely nothing out of it
", "rawMarkdown": "Update: Tried it, it didn't work. Could not make the submission work no matter what. Extremely disappointing as my first kaggle competition. I got absolutely nothing out of it", "votes": null } ], "comments": [ { "id": 3210501, "author_name": "quailwwk", "author_url": "", "post_date": "05/27/2025 09:35:41", "content": "I'm experiencing a similar submission problem. I've verified that the sample_submission.csv goes through the scoring procedure, and my own submission CSV also works correctly with the official scoring function by comparing it against the sample_submission.csv. But the formal submission still failed. Since I noticed you have a successful ranking, I was hoping you might have overcome this issue and could offer some advice?
after prediction use sample_submission.csv to match the rows
", "votes": null, "replies": [ { "id": 3210613, "author_name": "quailwwk", "author_url": "", "post_date": "05/27/2025 12:51:51", "content": "Thanks for the tip. I've tried it, but the same error persists.
if you use this \"Using to_csv(output_filename, index=False, mode='a', header=(i == 0)) to save submission rows can cause issues if the model crashes mid-way, resulting in an incomplete submission file. To avoid this, wrap the save logic in a try-except block so the error is skipped and at end use sample _submisison.csv to fine check where the skipped part will be replace and hence you get complete csv
", "votes": null, "replies": [ { "id": 3210768, "author_name": "kudos22", "author_url": "", "post_date": "05/27/2025 16:56:59", "content": "Well mine says a scoring error, so i imagine there's something up with the formatting of the file rather than the model crashing mid way. Still not sure where its going wrong, the formatting looks all the same to me
", "votes": null, "replies": [ { "id": 3210823, "author_name": "arunodhayan", "author_url": "", "post_date": "05/27/2025 17:48:06", "content": "you have a different testset , not the one available publicly, so if you script works well on public testset because the sequence length is not more than 720 , but in private testset it might be longer …
", "votes": null, "replies": [ { "id": 3210841, "author_name": "quailwwk", "author_url": "", "post_date": "05/27/2025 18:12:27", "content": "That makes sense, and yes, I got a successful submission by setting a length limit. But the weird thing is, if I just submit the plain sample_submission.csv (only containing the public testset, the notebook did not run any predictions), it still goes through the scoring process just fine.
Same, i also have the same dimensions of prediction dataset, doesn't seem to work
", "votes": null, "replies": [] } ] } ] }, { "id": 3210760, "author_name": "kudos22", "author_url": "", "post_date": "05/27/2025 16:49:22", "content": "I have a successful ranking because since nothing worked, i just submitted the sample_submission file as is, with 0s in all the columns. I still haven't been able to solve it. If you have, please let me know
", "votes": null, "replies": [ { "id": 3210767, "author_name": "quailwwk", "author_url": "", "post_date": "05/27/2025 16:56:32", "content": "I've only had one successful submission so far, and that was by skipping predictions for targets longer than 300 residues (R1126, R1136, R1138 – I used 'all zeros' for them), with no other modifications. If I try to include the predictions for these three, which my model can generate without issue, it still causes an error. It's quite annoying, as the final score is an overall average across all targets.
", "votes": null, "replies": [ { "id": 3210774, "author_name": "kudos22", "author_url": "", "post_date": "05/27/2025 16:59:45", "content": "That's honestly really weird and i'm not sure if its your fault tbh, let me try the same thing and see if i get a scoring error. This is my first kaggle competition and this has me in low energy tbh. Would be a shame if there's some issue with the scoring system itself if certain target ID's don't take predictions at all.
", "votes": null, "replies": [] }, { "id": 3212064, "author_name": "kudos22", "author_url": "", "post_date": "05/29/2025 09:01:44", "content": "Update: Tried it, it didn't work. Could not make the submission work no matter what. Extremely disappointing as my first kaggle competition. I got absolutely nothing out of it
", "votes": null, "replies": [] } ] } ] } ] } ], "raw_markdown_by_id": { "3206280": "It's my first time participating and I keep getting the 'submission scoring error' Which I think is due to invalid file format of my output compared to sample_submission. However I mannually and through code compared both the documents and couldn't seem to find any issues, same columns, rows, no header errors, spelling mistakes, encoding issues (utf-8), same datatypes for every column, unnecessary spaces; just replaced with my prediction data. Same index, spelling of headers etc. \n\nAfter multiple attempts I couldn't make it work. So after listening to someone's advice I just uploaded and submitted the sample_submission file with all zeros in the prediction columns. And that gave me a score. So I then took the sample_submission file and edited it to replace one of the zero columns (x_1) with both -1 and 1 (still float64 dtype) and then submitted again. And both of those failed. So I think something is wrong with placing numbers on the columns or something? Please help me how I can fix this scoring issue.", "3210501": "I'm experiencing a similar submission problem. I've verified that the `sample_submission.csv` goes through the scoring procedure, and my own submission CSV also works correctly with the official [scoring function](https://www.kaggle.com/code/metric/ribonanza-tm-score/notebook) by comparing it against the `sample_submission.csv`. But the formal submission still failed. Since I noticed you have a successful ranking, I was hoping you might have overcome this issue and could offer some advice?", "3210520": "after prediction use sample_submission.csv to match the rows", "3210613": "Thanks for the tip. I've tried it, but the same error persists. ", "3210691": "if you use this \"Using to_csv(output_filename, index=False, mode='a', header=(i == 0)) to save submission rows can cause issues if the model crashes mid-way, resulting in an incomplete submission file. To avoid this, wrap the save logic in a try-except block so the error is skipped and at end use sample _submisison.csv to fine check where the skipped part will be replace and hence you get complete csv", "3210760": "I have a successful ranking because since nothing worked, i just submitted the sample_submission file as is, with 0s in all the columns. I still haven't been able to solve it. If you have, please let me know", "3210761": "Same, i also have the same dimensions of prediction dataset, doesn't seem to work", "3210767": "I've only had one successful submission so far, and that was by skipping predictions for targets longer than 300 residues (R1126, R1136, R1138 – I used 'all zeros' for them), with no other modifications. If I try to include the predictions for these three, which my model can generate without issue, it still causes an error. It's quite annoying, as the final score is an overall average across all targets.", "3210768": "Well mine says a scoring error, so i imagine there's something up with the formatting of the file rather than the model crashing mid way. Still not sure where its going wrong, the formatting looks all the same to me", "3210774": "That's honestly really weird and i'm not sure if its your fault tbh, let me try the same thing and see if i get a scoring error. This is my first kaggle competition and this has me in low energy tbh. Would be a shame if there's some issue with the scoring system itself if certain target ID's don't take predictions at all.", "3210823": "you have a different testset , not the one available publicly, so if you script works well on public testset because the sequence length is not more than 720 , but in private testset it might be longer ...", "3210841": "That makes sense, and yes, I got a successful submission by setting a length limit. But the weird thing is, if I just submit the plain `sample_submission.csv` (only containing the public testset, the notebook did not run any predictions), it still goes through the scoring process just fine.", "3212064": "Update: Tried it, it didn't work. Could not make the submission work no matter what. Extremely disappointing as my first kaggle competition. I got absolutely nothing out of it" }, "source": "meta" }