I created a CNN model with pytorch and submitted it, but I get the error \"Notebook threw exception.\"
\nI would like to refer to the code of other participants, but I cannot find any notebooks that other participants have successfully submitted using test_images, so I am in a hopeless situation.
\nCan you give me some hints on how to solve this?
I created a CNN model with pytorch and submitted it, but I get the error \"Notebook threw exception.\"
\nI would like to refer to the code of other participants, but I cannot find any notebooks that other participants have successfully submitted using test_images, so I am in a hopeless situation.
\nCan you give me some hints on how to solve this?
I think you can submit it for the time being by using this code.
\n = pd.read_csv()\n[]=[int(i.()[0]) i [].values]\nsample_study_id = [].unique()\n = pd.read_csv()\n = [[]==].reset_index(=True)\ntest_stusy = [].unique()\n = .groupby()[].first().reset_index()\n\ntest_preds = inf_func(model, ) #model's output:25*3 cls\n\nnew_df = pd.DataFrame()\ntra_df = (pd.read_csv().columns[1:])\n\ncol = []\n i test_stusy:\n j tra_df:\n col.(f)\n\nnew_df[]=col\nnew_df[] = new_df[].astype()\nnew_df[]=0\nnew_df[]=0\nnew_df[]=0\nnew_df[[,,]]=test_preds.(-1,3)\n\n\nwithout = (sample_study_id)-(test_stusy)\n = pd.DataFrame()\n\n len(without)>0:\n col = []\n i without:\n j tra_df:\n col.(f)\n\n []=col\n [] = [].astype()\n []=0.34\n []=0.33\n []=0.33\n ()\n new_df = pd.concat([new_df,],axis=0).sort_values().reset_index(=True)\n new_df\nnew_df.to_csv(,index=False)\nHave you tried testing your inference pipeline with some examples from the training set, this way you can check how it performs with different study_id
", "rawMarkdown": "Have you tried testing your inference pipeline with some examples from the training set, this way you can check how it performs with different study_id", "votes": null }, { "id": "2859259", "postDate": "06/06/2024 22:26:48", "content": "I have the same error before, then I try to make a similar submission on trainingset, Inoticed that it is not always have full {instance_number}.dcm in every series, like 7.dcm, 9.dcm, 10.dcm… It miss 8.dcm, when I fix this, I can get a valid submission.
There is a starter notebook for PyTorch which is currently working: https://www.kaggle.com/code/shubhamcodez/rsna-resnet-starter-notebook
\nYou can compare and see if it fits your code.
@coderrkj Thank you for letting me know! I'll give it a try🥳
", "rawMarkdown": "coderrkj Thank you for letting me know! I'll give it a try🥳", "votes": null }, { "id": "2859552", "postDate": "06/07/2024 05:17:10", "content": "I see, thank you!
\nHowever, I use 'glob' to get all the dcm images, so that method doesn't seem to work.😭
@patriot Thanks for the sample code! I'll give it a try.
", "rawMarkdown": "patriot Thanks for the sample code! I'll give it a try.", "votes": null }, { "id": "2893568", "postDate": "06/27/2024 21:51:43", "content": "Were you able to identify what was causing the error? I'm facing the same problem…
", "rawMarkdown": "Were you able to identify what was causing the error? I'm facing the same problem...", "votes": null } ], "comments": [ { "id": 2854465, "author_name": "abebe9849", "author_url": "", "post_date": "06/04/2024 10:33:58", "content": "I think you can submit it for the time being by using this code.
\n = pd.read_csv()\n[]=[int(i.()[0]) i [].values]\nsample_study_id = [].unique()\n = pd.read_csv()\n = [[]==].reset_index(=True)\ntest_stusy = [].unique()\n = .groupby()[].first().reset_index()\n\ntest_preds = inf_func(model, ) #model's output:25*3 cls\n\nnew_df = pd.DataFrame()\ntra_df = (pd.read_csv().columns[1:])\n\ncol = []\n i test_stusy:\n j tra_df:\n col.(f)\n\nnew_df[]=col\nnew_df[] = new_df[].astype()\nnew_df[]=0\nnew_df[]=0\nnew_df[]=0\nnew_df[[,,]]=test_preds.(-1,3)\n\n\nwithout = (sample_study_id)-(test_stusy)\n = pd.DataFrame()\n\n len(without)>0:\n col = []\n i without:\n j tra_df:\n col.(f)\n\n []=col\n [] = [].astype()\n []=0.34\n []=0.33\n []=0.33\n ()\n new_df = pd.concat([new_df,],axis=0).sort_values().reset_index(=True)\n new_df\nnew_df.to_csv(,index=False)\n@patriot Thanks for the sample code! I'll give it a try.
", "votes": null, "replies": [] } ] }, { "id": 2857985, "author_name": "samu2505", "author_url": "", "post_date": "06/06/2024 08:08:49", "content": "Have you tried testing your inference pipeline with some examples from the training set, this way you can check how it performs with different study_id
", "votes": null, "replies": [] }, { "id": 2859259, "author_name": "athrunzala", "author_url": "", "post_date": "06/06/2024 22:26:48", "content": "I have the same error before, then I try to make a similar submission on trainingset, Inoticed that it is not always have full {instance_number}.dcm in every series, like 7.dcm, 9.dcm, 10.dcm… It miss 8.dcm, when I fix this, I can get a valid submission.
I see, thank you!
\nHowever, I use 'glob' to get all the dcm images, so that method doesn't seem to work.😭
There is a starter notebook for PyTorch which is currently working: https://www.kaggle.com/code/shubhamcodez/rsna-resnet-starter-notebook
\nYou can compare and see if it fits your code.
@coderrkj Thank you for letting me know! I'll give it a try🥳
", "votes": null, "replies": [] } ] }, { "id": 2893568, "author_name": "hmssantos", "author_url": "", "post_date": "06/27/2024 21:51:43", "content": "Were you able to identify what was causing the error? I'm facing the same problem…
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2852305": "I created a CNN model with pytorch and submitted it, but I get the error \"Notebook threw exception.\"\nI would like to refer to the code of other participants, but I cannot find any notebooks that other participants have successfully submitted using test_images, so I am in a hopeless situation.\nCan you give me some hints on how to solve this?", "2854465": "I think you can submit it for the time being by using this code.\n\n```\nsample = pd.read_csv(\"/kaggle/input/rsna-2024-lumbar-spine-degenerative-classification/sample_submission.csv\")\nsample[\"study_id\"]=[int(i.split(\"_\")[0]) for i in sample[\"row_id\"].values]\nsample_study_id = sample[\"study_id\"].unique()\ntest = pd.read_csv(\"/kaggle/input/rsna-2024-lumbar-spine-degenerative-classification/test_series_descriptions.csv\")\ntest = test[test[\"series_description\"]==\"Axial T2\"].reset_index(drop=True)\ntest_stusy = test[\"study_id\"].unique()\ntest = test.groupby(\"study_id\")[\"series_id\"].first().reset_index()\n\ntest_preds = inf_func(model, test) #model's output:25*3 cls\n\nnew_df = pd.DataFrame()\ntra_df = list(pd.read_csv(\"/kaggle/input/rsna-2024-lumbar-spine-degenerative-classification/train.csv\").columns[1:])\n\ncol = []\nfor i in test_stusy:\n for j in tra_df:\n col.append(f\"{i}_{j}\")\n \nnew_df[\"row_id\"]=col\nnew_df[\"row_id\"] = new_df[\"row_id\"].astype(\"str\")\nnew_df[\"normal_mild\"]=0\nnew_df[\"moderate\"]=0\nnew_df[\"severe\"]=0\nnew_df[[\"normal_mild\",\"moderate\",\"severe\"]]=test_preds.reshape(-1,3)\n\n\nwithout = set(sample_study_id)-set(test_stusy)\nsample = pd.DataFrame()\n\nif len(without)>0:\n col = []\n for i in without:\n for j in tra_df:\n col.append(f\"{i}_{j}\")\n\n sample[\"row_id\"]=col\n sample[\"row_id\"] = sample[\"row_id\"].astype(\"str\")\n sample[\"normal_mild\"]=0.34\n sample[\"moderate\"]=0.33\n sample[\"severe\"]=0.33\n print(sample)\n new_df = pd.concat([new_df,sample],axis=0).sort_values(\"row_id\").reset_index(drop=True)\n new_df\nnew_df.to_csv(\"submission.csv\",index=False)\n```", "2857985": "Have you tried testing your inference pipeline with some examples from the training set, this way you can check how it performs with different study_id", "2859259": "I have the same error before, then I try to make a similar submission on trainingset, Inoticed that it is not always have full {instance_number}.dcm in every series, like `7.dcm`, `9.dcm`, `10.dcm`... It miss `8.dcm`, when I fix this, I can get a valid submission.", "2859436": "There is a starter notebook for PyTorch which is currently working: https://www.kaggle.com/code/shubhamcodez/rsna-resnet-starter-notebook\nYou can compare and see if it fits your code.", "2859548": "coderrkj Thank you for letting me know! I'll give it a try🥳", "2859552": "I see, thank you!\nHowever, I use 'glob' to get all the dcm images, so that method doesn't seem to work.😭", "2859554": "patriot Thanks for the sample code! I'll give it a try.", "2893568": "Were you able to identify what was causing the error? I'm facing the same problem..." }, "source": "meta" }