I've tried some different models, but every time my score is 0.0614. I even make all the predictions -1 and submit, it was expected to get 0 score, but it is still 0.0614.Who give me some possibilities about what happened? Thanks to you all.
", "messages": [ { "id": "761104", "postDate": "03/02/2020 06:20:51", "content": "I've tried some different models, but every time my score is 0.0614. I even make all the predictions -1 and submit, it was expected to get 0 score, but it is still 0.0614.Who give me some possibilities about what happened? Thanks to you all.
", "rawMarkdown": "I've tried some different models, but every time my score is 0.0614. I even make all the predictions -1 and submit, it was expected to get 0 score, but it is still 0.0614.Who give me some possibilities about what happened? Thanks to you all.", "votes": null }, { "id": "761284", "postDate": "03/02/2020 11:03:32", "content": "Have you tried running your inference code on the train data set and checked the outputs ? I would recommend using a starter inference kernel and switching their model with yours.
", "rawMarkdown": "Have you tried running your inference code on the train data set and checked the outputs ? I would recommend using a starter inference kernel and switching their model with yours.", "votes": null }, { "id": "761286", "postDate": "03/02/2020 11:05:19", "content": "That is the baseline score when submitting the original csv, so probably you are not saving your csv correctly or something happens in similar manner.
", "rawMarkdown": "That is the baseline score when submitting the original csv, so probably you are not saving your csv correctly or something happens in similar manner.", "votes": null }, { "id": "761305", "postDate": "03/02/2020 11:25:50", "content": "Thanks to you all,I think I've found the solvement.I just modify the sample_submission.csv which contains only 36 rows and then save it to the working content.May be write the predictions to a new csv file can solve it.But it is still strange that the score based on the modified sample_submission.csv is the same as based on the original file,I can sure that I've saved the output file correctly, can anyone give some explanation?
", "rawMarkdown": "Thanks to you all,I think I've found the solvement.I just modify the sample_submission.csv which contains only 36 rows and then save it to the working content.May be write the predictions to a new csv file can solve it.But it is still strange that the score based on the modified sample_submission.csv is the same as based on the original file,I can sure that I've saved the output file correctly, can anyone give some explanation?", "votes": null }, { "id": "761401", "postDate": "03/02/2020 13:23:35", "content": "It is a good idea to create the submission file by reading the first column of the test parquet files or the test csv file.
", "rawMarkdown": "It is a good idea to create the submission file by reading the first column of the test parquet files or the test csv file.", "votes": null } ], "comments": [ { "id": 761284, "author_name": "utsavnandi", "author_url": "", "post_date": "03/02/2020 11:03:32", "content": "Have you tried running your inference code on the train data set and checked the outputs ? I would recommend using a starter inference kernel and switching their model with yours.
", "votes": null, "replies": [] }, { "id": 761286, "author_name": "philippsinger", "author_url": "", "post_date": "03/02/2020 11:05:19", "content": "That is the baseline score when submitting the original csv, so probably you are not saving your csv correctly or something happens in similar manner.
", "votes": null, "replies": [] }, { "id": 761305, "author_name": "huyunhui", "author_url": "", "post_date": "03/02/2020 11:25:50", "content": "Thanks to you all,I think I've found the solvement.I just modify the sample_submission.csv which contains only 36 rows and then save it to the working content.May be write the predictions to a new csv file can solve it.But it is still strange that the score based on the modified sample_submission.csv is the same as based on the original file,I can sure that I've saved the output file correctly, can anyone give some explanation?
", "votes": null, "replies": [ { "id": 761401, "author_name": "utsavnandi", "author_url": "", "post_date": "03/02/2020 13:23:35", "content": "It is a good idea to create the submission file by reading the first column of the test parquet files or the test csv file.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "761104": "I've tried some different models, but every time my score is 0.0614. I even make all the predictions -1 and submit, it was expected to get 0 score, but it is still 0.0614.Who give me some possibilities about what happened? Thanks to you all.", "761284": "Have you tried running your inference code on the train data set and checked the outputs ? I would recommend using a starter inference kernel and switching their model with yours.", "761286": "That is the baseline score when submitting the original csv, so probably you are not saving your csv correctly or something happens in similar manner.", "761305": "Thanks to you all,I think I've found the solvement.I just modify the sample_submission.csv which contains only 36 rows and then save it to the working content.May be write the predictions to a new csv file can solve it.But it is still strange that the score based on the modified sample_submission.csv is the same as based on the original file,I can sure that I've saved the output file correctly, can anyone give some explanation?", "761401": "It is a good idea to create the submission file by reading the first column of the test parquet files or the test csv file." }, "source": "meta" }