Hi, our team is really struggling with finding a viable way to submit the solution. This is our current notebook, is anyone able to look over it to provide some suggestions? We have implemented as much as we can from the previous threads to no avail, hence we are starting a new thread.
\nhttps://www.kaggle.com/hkrsmk/inference-yolov5-hkrsmk
\nAny help would be greatly appreciated.
\nEdit - we found the solution: don't output any CSV files manually, let the API do it on its own.
", "messages": [ { "id": 1637248, "postDate": "2022-01-03T17:27:18.983Z", "content": "Hi, our team is really struggling with finding a viable way to submit the solution. This is our current notebook, is anyone able to look over it to provide some suggestions? We have implemented as much as we can from the previous threads to no avail, hence we are starting a new thread.
\nhttps://www.kaggle.com/hkrsmk/inference-yolov5-hkrsmk
\nAny help would be greatly appreciated.
\nEdit - we found the solution: don't output any CSV files manually, let the API do it on its own.
", "rawMarkdown": "Hi, our team is really struggling with finding a viable way to submit the solution. This is our current notebook, is anyone able to look over it to provide some suggestions? We have implemented as much as we can from the previous threads to no avail, hence we are starting a new thread.\n\nhttps://www.kaggle.com/hkrsmk/inference-yolov5-hkrsmk\n\nAny help would be greatly appreciated.\n\nEdit - we found the solution: don't output any CSV files manually, let the API do it on its own.", "votes": 3 }, { "id": 1642062, "postDate": "2022-01-08T02:24:22.810Z", "content": "Solved the issue - the API generates a submission csv when we run it, we do not need to create a submission csv ourselves elsewhere. So we just remove the df.to_csv part.
I think you are overriding index too, how about just registering prediction instead of index too. have you tried? 😺
", "rawMarkdown": "I think you are overriding index too, how about just registering prediction instead of index too. have you tried? 😺", "replies": [ { "id": 1637848, "postDate": "2022-01-04T08:56:05.317Z", "content": "You mean submission_dict['index'].append(idc)? We are adding to a different dictionary for visibility; otherwise, we do not modify sample_prediction_df['index'] at all, and we submit the second dataframe for the competition:
\nenv.predict(sample_prediction_df)
I mean that instead of passing the second df, I want you to try only giving predictions as per the competition's evaluation page, without altering the index. I'm thinking that error, might be because of, passing two columns (index, preds) , while it may be expecting only preds columns. if you are still stuck on this error. I want you to try like this ; )
", "rawMarkdown": "I mean that instead of passing the second df, I want you to try only giving predictions as per the competition's evaluation page, without altering the index. I'm thinking that error, might be because of, passing two columns (index, preds) , while it may be expecting only preds columns. if you are still stuck on this error. I want you to try like this ; )" } ] }, { "id": 1637580, "postDate": "2022-01-04T03:26:27.090Z", "content": "@addisonhoward Any thoughts, or ways we can contact the organisers for clarification?
", "rawMarkdown": "@addisonhoward Any thoughts, or ways we can contact the organisers for clarification?", "replies": [ { "id": 1637594, "postDate": "2022-01-04T03:57:45.027Z", "content": "Hi - unfortunately neither we at Kaggle or the host can troubleshoot code that is otherwise working for other teams. If we notice a site-wide error, we can investigate.
", "rawMarkdown": "Hi - unfortunately neither we at Kaggle or the host can troubleshoot code that is otherwise working for other teams. If we notice a site-wide error, we can investigate. " }, { "id": 1637606, "postDate": "2022-01-04T04:13:09.207Z", "content": "Cool thanks for the quick reply!
", "rawMarkdown": "Cool thanks for the quick reply!" } ] }, { "id": 1637253, "postDate": "2022-01-03T17:34:18.513Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 1637269, "postDate": "2022-01-03T17:53:35.950Z", "content": "I saw some teams having submission times of 1h, so maybe not?
", "rawMarkdown": "I saw some teams having submission times of 1h, so maybe not?" } ] } ], "comments": [ { "id": 1642062, "author_name": "hkrsmk", "author_url": "", "post_date": "2022-01-08T02:24:22.810000", "content": "Solved the issue - the API generates a submission csv when we run it, we do not need to create a submission csv ourselves elsewhere. So we just remove the df.to_csv part.
I think you are overriding index too, how about just registering prediction instead of index too. have you tried? 😺
", "votes": 0, "replies": [ { "id": 1637848, "author_name": "hkrsmk", "author_url": "", "post_date": "2022-01-04T08:56:05.317000", "content": "You mean submission_dict['index'].append(idc)? We are adding to a different dictionary for visibility; otherwise, we do not modify sample_prediction_df['index'] at all, and we submit the second dataframe for the competition:
\nenv.predict(sample_prediction_df)
I mean that instead of passing the second df, I want you to try only giving predictions as per the competition's evaluation page, without altering the index. I'm thinking that error, might be because of, passing two columns (index, preds) , while it may be expecting only preds columns. if you are still stuck on this error. I want you to try like this ; )
", "votes": 0, "replies": [] } ] }, { "id": 1637580, "author_name": "hkrsmk", "author_url": "", "post_date": "2022-01-04T03:26:27.090000", "content": "@addisonhoward Any thoughts, or ways we can contact the organisers for clarification?
", "votes": 0, "replies": [ { "id": 1637594, "author_name": "Addison Howard", "author_url": "", "post_date": "2022-01-04T03:57:45.027000", "content": "Hi - unfortunately neither we at Kaggle or the host can troubleshoot code that is otherwise working for other teams. If we notice a site-wide error, we can investigate.
", "votes": 0, "replies": [] }, { "id": 1637606, "author_name": "hkrsmk", "author_url": "", "post_date": "2022-01-04T04:13:09.207000", "content": "Cool thanks for the quick reply!
", "votes": 0, "replies": [] } ] }, { "id": 1637253, "author_name": "", "author_url": "", "post_date": "2022-01-03T17:34:18.513000", "content": "", "votes": 0, "replies": [ { "id": 1637269, "author_name": "hkrsmk", "author_url": "", "post_date": "2022-01-03T17:53:35.950000", "content": "I saw some teams having submission times of 1h, so maybe not?
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1637248": "Hi, our team is really struggling with finding a viable way to submit the solution. This is our current notebook, is anyone able to look over it to provide some suggestions? We have implemented as much as we can from the previous threads to no avail, hence we are starting a new thread.\n\nhttps://www.kaggle.com/hkrsmk/inference-yolov5-hkrsmk\n\nAny help would be greatly appreciated.\n\nEdit - we found the solution: don't output any CSV files manually, let the API do it on its own.", "1642062": "Solved the issue - the API generates a submission csv when we run it, we do not need to create a submission csv ourselves elsewhere. So we just remove the `df.to_csv` part.", "1637714": "I think you are overriding index too, how about just registering prediction instead of index too. have you tried? 😺", "1637580": "@addisonhoward Any thoughts, or ways we can contact the organisers for clarification?", "1637253": "" } }