![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F3459992%2F4ee0b59727315f187dbeb8b8c272e09b%2Funknown.png?generation=1588098856293436&alt=media\")
I tried making a submission, but got an error saying submission.csv not found , while it exists.
\n\n
I used the inference part from this kernel
\n\nAnyone having similar issues / or know a solution?\nThank you.
\n\nEDIT - There was an issue in the inference code. The inference function was being passed a parameter for pretrained , and there was a typo in the variable name.
", "messages": [ { "id": 859021, "postDate": "2020-05-24T04:12:59.173Z", "content": "I got the same issue as well. The submission csv is sometimes found by kaggle and sometimes not.\nWhat i meant was ,2 of my submissions went through successfull. However all the later submissions with only changes in model and nothing else fails to generate the csv. \nI wonder whats the real issue here. I have tried reducing file size, reducing batch size, almost everything.\nAny help is appreciated.
", "rawMarkdown": "I got the same issue as well. The submission csv is sometimes found by kaggle and sometimes not.\nWhat i meant was ,2 of my submissions went through successfull. However all the later submissions with only changes in model and nothing else fails to generate the csv. \nI wonder whats the real issue here. I have tried reducing file size, reducing batch size, almost everything.\nAny help is appreciated.", "replies": [ { "id": 860846, "postDate": "2020-05-25T16:20:41.427Z", "content": "It's probably because the file is not actually being created when you run the inference on the complete public test set , due to some incorrect file naming , function calls or things like that. You said you only changed the model , which I assume you are creating an instance of , or loading the weights in the inference function or loop , so there might be some error there in loading the model , the weights or anything like that which might cause the csv file to not be generated during the submission. A simple workaround is to just try the inference code on the training files , just load the train.csv and run inference for train.head() and if there is any error you will see the traceback for it.
", "rawMarkdown": "It's probably because the file is not actually being created when you run the inference on the complete public test set , due to some incorrect file naming , function calls or things like that. You said you only changed the model , which I assume you are creating an instance of , or loading the weights in the inference function or loop , so there might be some error there in loading the model , the weights or anything like that which might cause the csv file to not be generated during the submission. A simple workaround is to just try the inference code on the training files , just load the train.csv and run inference for train.head() and if there is any error you will see the traceback for it. " } ] }, { "id": 847018, "postDate": "2020-05-14T06:01:00.937Z", "content": "@p4rallax \nHi i get the same issue how did u overcome the issue.
", "rawMarkdown": "@p4rallax \nHi i get the same issue how did u overcome the issue. \n", "replies": [ { "id": 847816, "postDate": "2020-05-14T16:03:13.563Z", "content": "There was an issue with the code I used for inference. The inference function was not being called properly. Check your code again , where you create the submission.csv. Make sure you are able to run the inference code on any number of images and not just the 3 as in the test.csv provided. Try checking the inference code on the training images.
", "rawMarkdown": "There was an issue with the code I used for inference. The inference function was not being called properly. Check your code again , where you create the submission.csv. Make sure you are able to run the inference code on any number of images and not just the 3 as in the test.csv provided. Try checking the inference code on the training images." } ] }, { "id": 825043, "postDate": "2020-04-28T18:35:59.960Z", "content": "I tried making a submission, but got an error saying submission.csv not found , while it exists.
\n\n
I used the inference part from this kernel
\n\nAnyone having similar issues / or know a solution?\nThank you.
\n\nEDIT - There was an issue in the inference code. The inference function was being passed a parameter for pretrained , and there was a typo in the variable name.
", "rawMarkdown": "I tried making a submission, but got an error saying submission.csv not found , while it exists.\n\n\n\nI used the inference part from [this kernel](https://www.kaggle.com/yasufuminakama/panda-se-resnext50-classification-baseline)\n\nAnyone having similar issues / or know a solution?\nThank you.\n\n\n\nEDIT - There was an issue in the inference code. The inference function was being passed a parameter for pretrained , and there was a typo in the variable name. " } ], "comments": [ { "id": 859021, "author_name": "Noble", "author_url": "", "post_date": "2020-05-24T04:12:59.173000", "content": "I got the same issue as well. The submission csv is sometimes found by kaggle and sometimes not.\nWhat i meant was ,2 of my submissions went through successfull. However all the later submissions with only changes in model and nothing else fails to generate the csv. \nI wonder whats the real issue here. I have tried reducing file size, reducing batch size, almost everything.\nAny help is appreciated.
", "votes": 0, "replies": [ { "id": 860846, "author_name": "Satwik", "author_url": "", "post_date": "2020-05-25T16:20:41.427000", "content": "It's probably because the file is not actually being created when you run the inference on the complete public test set , due to some incorrect file naming , function calls or things like that. You said you only changed the model , which I assume you are creating an instance of , or loading the weights in the inference function or loop , so there might be some error there in loading the model , the weights or anything like that which might cause the csv file to not be generated during the submission. A simple workaround is to just try the inference code on the training files , just load the train.csv and run inference for train.head() and if there is any error you will see the traceback for it.
", "votes": 0, "replies": [] } ] }, { "id": 847018, "author_name": "Jaideep", "author_url": "", "post_date": "2020-05-14T06:01:00.937000", "content": "@p4rallax \nHi i get the same issue how did u overcome the issue.
", "votes": 0, "replies": [ { "id": 847816, "author_name": "Satwik", "author_url": "", "post_date": "2020-05-14T16:03:13.563000", "content": "There was an issue with the code I used for inference. The inference function was not being called properly. Check your code again , where you create the submission.csv. Make sure you are able to run the inference code on any number of images and not just the 3 as in the test.csv provided. Try checking the inference code on the training images.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "859021": "I got the same issue as well. The submission csv is sometimes found by kaggle and sometimes not.\nWhat i meant was ,2 of my submissions went through successfull. However all the later submissions with only changes in model and nothing else fails to generate the csv. \nI wonder whats the real issue here. I have tried reducing file size, reducing batch size, almost everything.\nAny help is appreciated.", "847018": "@p4rallax \nHi i get the same issue how did u overcome the issue. \n", "825043": "I tried making a submission, but got an error saying submission.csv not found , while it exists.\n\n\n\nI used the inference part from [this kernel](https://www.kaggle.com/yasufuminakama/panda-se-resnext50-classification-baseline)\n\nAnyone having similar issues / or know a solution?\nThank you.\n\n\n\nEDIT - There was an issue in the inference code. The inference function was being passed a parameter for pretrained , and there was a typo in the variable name. " } }