![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F846065%2F3bbae72b1ec377fc24ed4407b61c12ba%2FScreenshot%20from%202020-06-28%2014-19-46.png?generation=1593368446651864&alt=media\")
New to machine learning and data science? No question is too basic or too simple. Feel free to start your own thread, or use this thread as a place to post any first-timer clarifying questions for the Kaggle community to help you with!
\n\nIf you would consider yourself a beginner but don't know where to get started, let other Kagglers help you take your first steps here!
\n\nNew to Kaggle? Take a look at a few videos our very own Dr. Rachael Tatman has put together to learn a bit more about site etiquette, Kaggle lingo, and how to enter a competition using Kaggle Notebooks.
", "messages": [ { "id": "833429", "postDate": "05/04/2020 19:56:07", "content": "New to machine learning and data science? No question is too basic or too simple. Feel free to start your own thread, or use this thread as a place to post any first-timer clarifying questions for the Kaggle community to help you with!
\n\nIf you would consider yourself a beginner but don't know where to get started, let other Kagglers help you take your first steps here!
\n\nNew to Kaggle? Take a look at a few videos our very own Dr. Rachael Tatman has put together to learn a bit more about site etiquette, Kaggle lingo, and how to enter a competition using Kaggle Notebooks.
", "rawMarkdown": "New to machine learning and data science? No question is too basic or too simple. Feel free to start your own thread, or use this thread as a place to post any first-timer clarifying questions for the Kaggle community to help you with! \n\nIf you would consider yourself a beginner but don't know where to get started, let other Kagglers help you take your first steps here!\n\nNew to Kaggle? Take a look at a few videos our very own Dr. Rachael Tatman has put together to learn a bit more about [site etiquette](https://www.youtube.com/watch?v=aIus8si_Et0), [Kaggle lingo](https://www.youtube.com/watch?v=sEJHyuWKd-s), and [how to enter a competition using Kaggle Notebooks](https://www.youtube.com/watch?&v=GJBOMWpLpTQ).", "votes": null }, { "id": "834790", "postDate": "05/05/2020 19:19:44", "content": "Hi, \n I have just started doing basic ML stuff and this is my first time on kaggle.\nI am trying to implement a simple Neural Network. Is the training data given to us or do we have to find it ourselves and then use the add data button in the data section to use it?\nIs the input file empty given to us in the data section empty?\nThanks
", "rawMarkdown": "Hi, \n I have just started doing basic ML stuff and this is my first time on kaggle.\nI am trying to implement a simple Neural Network. Is the training data given to us or do we have to find it ourselves and then use the add data button in the data section to use it?\nIs the input file empty given to us in the data section empty?\nThanks", "votes": null }, { "id": "834921", "postDate": "05/05/2020 22:26:34", "content": "We have been given the training data.
", "rawMarkdown": "We have been given the training data.", "votes": null }, { "id": "835066", "postDate": "05/06/2020 03:02:10", "content": "Hi Kanishk,\nsuggest you to go through this.\nKaggle Tutorial - Titanic Competition
", "rawMarkdown": "Hi Kanishk,\nsuggest you to go through this.\n[Kaggle Tutorial - Titanic Competition](https://www.kaggle.com/alexisbcook/titanic-tutorial)", "votes": null }, { "id": "835096", "postDate": "05/06/2020 03:44:12", "content": "Hi ,
\n\nI have just finished my basic ML course and first time in Kaggle. \nI have tried doing Titanic competition and could understand the process. It would be great if you could guide me?
\n\nThanks
", "rawMarkdown": "Hi ,\n\nI have just finished my basic ML course and first time in Kaggle. \nI have tried doing Titanic competition and could understand the process. It would be great if you could guide me?\n\nThanks", "votes": null }, { "id": "835242", "postDate": "05/06/2020 06:19:55", "content": "Happy to have you joining the competition !
\n\nYou can start by looking at these notebooks:\n- https://www.kaggle.com/bendvd/global-wheat-detection-first-submission\n- https://www.kaggle.com/pestipeti/pytorch-starter-fasterrcnn-train
", "rawMarkdown": "Happy to have you joining the competition !\n\nYou can start by looking at these notebooks:\n- https://www.kaggle.com/bendvd/global-wheat-detection-first-submission\n- https://www.kaggle.com/pestipeti/pytorch-starter-fasterrcnn-train", "votes": null }, { "id": "837883", "postDate": "05/08/2020 05:37:51", "content": "I am new to Kaggle and the submission process (or rules??) is not clear at all. The submit to competition button is disabled and mouse hover over it shows: \"You do not have a competition data source.\"\nCan someone explain what that means?
\n\nSimilar to the \"titanic tutorial\" etc linked below I am trying to fine tune a keras_retinanet model and then make predictions. I believe I have successfully created the submissions.csv file.
\n\nAll the tutorials and shared kernels seem to be using Pytorch. But I am trying to use this repo which basically provides pretrained models and framework for finetuning/inference just like Pytorch but with Keras/Tensorflow:
\n\nhttps://github.com/fizyr/keras-retinanet
\n\nI would be happy to share any more details or codes (training/testing anything) etc. if anyone can help me understand what's going on.
\n\nThanks!
", "rawMarkdown": "I am new to Kaggle and the submission process (or rules??) is not clear at all. The submit to competition button is disabled and mouse hover over it shows: \"You do not have a competition data source.\"\nCan someone explain what that means?\n\nSimilar to the \"titanic tutorial\" etc linked below I am trying to fine tune a keras_retinanet model and then make predictions. I believe I have successfully created the submissions.csv file.\n\nAll the tutorials and shared kernels seem to be using Pytorch. But I am trying to use this repo which basically provides pretrained models and framework for finetuning/inference just like Pytorch but with Keras/Tensorflow: \n\nhttps://github.com/fizyr/keras-retinanet\n\nI would be happy to share any more details or codes (training/testing anything) etc. if anyone can help me understand what's going on.\n\nThanks!", "votes": null }, { "id": "839011", "postDate": "05/09/2020 02:06:25", "content": "When I am turning off internet, the public pre-trained model is not getting downloaded and the script is failing. So that means we can't use transfer learning or there is a workaround ? This works fine with internet on. Please help.
", "rawMarkdown": "When I am turning off internet, the public pre-trained model is not getting downloaded and the script is failing. So that means we can't use transfer learning or there is a workaround ? This works fine with internet on. Please help.", "votes": null }, { "id": "839065", "postDate": "05/09/2020 03:30:11", "content": "I have seen people using two different notebook instances for training and inference. Internet off criteria is set for inference part only.\nYou can check those notebook on notebooks section.
", "rawMarkdown": "I have seen people using two different notebook instances for training and inference. Internet off criteria is set for inference part only.\nYou can check those notebook on notebooks section.", "votes": null }, { "id": "839085", "postDate": "05/09/2020 04:01:16", "content": "Thanks for your response.
\n\nSo we can share file between notebooks ? Unfortunately when I saved the version ( run and save ), I couldn't find the weights file in the output folder. I may need to dig in more.
", "rawMarkdown": "Thanks for your response.\n\nSo we can share file between notebooks ? Unfortunately when I saved the version ( run and save ), I couldn't find the weights file in the output folder. I may need to dig in more.", "votes": null }, { "id": "839089", "postDate": "05/09/2020 04:09:19", "content": "The general strategy here is to have two notebooks: one for training (with internet) and another to perform inference (w/o internet). To get your trained weights to your inference notebook, you should upload them as a Kaggle dataset which your notebook can have access to even with no internet.
\n\nAs Rahat said, there's a few public kernels that use this approach so you should take a look to see how they import the weights.
", "rawMarkdown": "The general strategy here is to have two notebooks: one for training (with internet) and another to perform inference (w/o internet). To get your trained weights to your inference notebook, you should upload them as a Kaggle dataset which your notebook can have access to even with no internet.\n\nAs Rahat said, there's a few public kernels that use this approach so you should take a look to see how they import the weights.", "votes": null }, { "id": "839099", "postDate": "05/09/2020 04:25:23", "content": "Appreciate your quick response Matthew. I was able to break the scripts and submit the score.
", "rawMarkdown": "Appreciate your quick response Matthew. I was able to break the scripts and submit the score.", "votes": null }, { "id": "852047", "postDate": "05/18/2020 06:18:15", "content": "Hi\nI am trying to compete in Global Wheat Detection\nIn the sample_submission there are five entries in the predictive strings for each image id.\nI am wondering what are those values?
", "rawMarkdown": "Hi\nI am trying to compete in Global Wheat Detection\nIn the sample_submission there are five entries in the predictive strings for each image id.\nI am wondering what are those values?", "votes": null }, { "id": "863307", "postDate": "05/27/2020 08:14:00", "content": "Here's my youtube blog companion videos on the competition, helps me be motivated. Might be of help to someone else to get started : https://youtu.be/zxT8_lyA6fc
", "rawMarkdown": "Here's my youtube blog companion videos on the competition, helps me be motivated. Might be of help to someone else to get started : https://youtu.be/zxT8_lyA6fc", "votes": null }, { "id": "904964", "postDate": "06/28/2020 05:52:13", "content": "I have created my notebook with submissions, but I am unable to find the submit button, I checked the documentation and followed all the rules but I am still unable to submit, can anyone help?
", "rawMarkdown": "I have created my notebook with submissions, but I am unable to find the submit button, I checked the documentation and followed all the rules but I am still unable to submit, can anyone help?", "votes": null }, { "id": "904977", "postDate": "06/28/2020 06:26:11", "content": "Commit your notebook. Then go to the viewer and look at the latest version. Scroll down to the output section and click on the 'submission.csv' you generated. Then there should be a submit button that you should see.
", "rawMarkdown": "Commit your notebook. Then go to the viewer and look at the latest version. Scroll down to the output section and click on the 'submission.csv' you generated. Then there should be a submit button that you should see.", "votes": null }, { "id": "905438", "postDate": "06/28/2020 14:50:01", "content": "I am unable to find this, is it possible for you to share a couple of screenshots with me?\nThis is what my csv looks like:\nhttps://imgur.com/a/QvoUGcd
", "rawMarkdown": "I am unable to find this, is it possible for you to share a couple of screenshots with me?\nThis is what my csv looks like:\nhttps://imgur.com/a/QvoUGcd", "votes": null }, { "id": "905444", "postDate": "06/28/2020 14:55:18", "content": "Hi everyone
\n\nHere are my two first notebook: \nhttps://www.kaggle.com/syphax93/interactive-visualization-modelling-0-803-top7 \nhttps://www.kaggle.com/syphax93/vapnik-s-theory-bias-variance-and-learning-curve
\n\nI'd like to have your opinion.Thanks
", "rawMarkdown": "Hi everyone \n\nHere are my two first notebook: \nhttps://www.kaggle.com/syphax93/interactive-visualization-modelling-0-803-top7 \nhttps://www.kaggle.com/syphax93/vapnik-s-theory-bias-variance-and-learning-curve \n\nI'd like to have your opinion.Thanks", "votes": null }, { "id": "905707", "postDate": "06/28/2020 18:20:47", "content": "I think it's because you didn't name it 'submission.csv'. Try again and it should look like this with a blue submit button:\n
Ok, I have tried changing the configuration of my file to \"Save output for this run\" and ranyt it, but I cannot find the output, anything that can be done?
\n\nAlso, sorry to bother you so much with the beginner questions.
", "rawMarkdown": "Ok, I have tried changing the configuration of my file to \"Save output for this run\" and ranyt it, but I cannot find the output, anything that can be done?\n\nAlso, sorry to bother you so much with the beginner questions.", "votes": null }, { "id": "905771", "postDate": "06/28/2020 19:21:05", "content": "Not sure what you mean by the \"Save output for this run\". Did you commit your notebook again?
", "rawMarkdown": "Not sure what you mean by the \"Save output for this run\". Did you commit your notebook again?", "votes": null }, { "id": "905778", "postDate": "06/28/2020 19:31:59", "content": "Yes, I did commit it again with a save for version configuration\nVersion Settings\nSave outputs when creating a Quick Save
\n\nNever save output
\n\nAlways save output
\n\nSave output for this version\nSave & Run All with an accelerator
\n\nWhen I opened advanced settings on the save button, this is what I used to save output for version.
", "rawMarkdown": "Yes, I did commit it again with a save for version configuration\nVersion Settings\nSave outputs when creating a Quick Save\n\n\nNever save output\n\nAlways save output\n\nSave output for this version\nSave & Run All with an accelerator\n\nWhen I opened advanced settings on the save button, this is what I used to save output for version.", "votes": null }, { "id": "905799", "postDate": "06/28/2020 20:04:08", "content": "Not really sure what you mean. So when you are editing your notebook, you pressed \"Save Version\" and then \"Save & Run All (Commit)\", right? Then you go to the viewer and look at that last version.
", "rawMarkdown": "Not really sure what you mean. So when you are editing your notebook, you pressed \"Save Version\" and then \"Save & Run All (Commit)\", right? Then you go to the viewer and look at that last version.", "votes": null }, { "id": "906018", "postDate": "06/29/2020 03:11:08", "content": "Yes, something like that:\nFirst I opened the advanced settings, and clicked on save output for this version.\nThen I clicked on Save & Run All.
", "rawMarkdown": "Yes, something like that:\nFirst I opened the advanced settings, and clicked on save output for this version.\nThen I clicked on Save & Run All.", "votes": null }, { "id": "906612", "postDate": "06/29/2020 12:51:14", "content": "And you're still having trouble finding the submit?
", "rawMarkdown": "And you're still having trouble finding the submit?", "votes": null }, { "id": "906689", "postDate": "06/29/2020 13:50:05", "content": "Yeah, I am not able to find the output file.
", "rawMarkdown": "Yeah, I am not able to find the output file.", "votes": null }, { "id": "906697", "postDate": "06/29/2020 13:53:15", "content": "Does it not generate or what? I'm sorry but I'm not sure how much further I can help. Just try to review what I have already told you
", "rawMarkdown": "Does it not generate or what? I'm sorry but I'm not sure how much further I can help. Just try to review what I have already told you", "votes": null }, { "id": "906698", "postDate": "06/29/2020 13:54:07", "content": "Ok, thanks for your help, Im just confused as to how to save the file once it opens...
", "rawMarkdown": "Ok, thanks for your help, Im just confused as to how to save the file once it opens...", "votes": null }, { "id": "906701", "postDate": "06/29/2020 13:56:06", "content": "Still I'm not sure what you mean.\n- In your notebook, within the code, you should save your predictions as 'submission.csv'.\n- Then commit.\n- Then view your latest commit in the viewer.\n- Scroll down to the output, like you showed me in the screenshot before.\n- Click on the submission.csv file.\n- And click submit.
", "rawMarkdown": "Still I'm not sure what you mean.\n- In your notebook, within the code, you should save your predictions as 'submission.csv'.\n- Then commit.\n- Then view your latest commit in the viewer.\n- Scroll down to the output, like you showed me in the screenshot before.\n- Click on the submission.csv file.\n- And click submit.", "votes": null }, { "id": "906703", "postDate": "06/29/2020 13:57:20", "content": "That was not my submission file, it was the sample submission file that they gave, I am not able to find my submission.csv in the viewer.
", "rawMarkdown": "That was not my submission file, it was the sample submission file that they gave, I am not able to find my submission.csv in the viewer.", "votes": null }, { "id": "906712", "postDate": "06/29/2020 14:01:13", "content": "Then perhaps the portion of code that saves your submission is not getting run? Check above in your code to see if it all got ran/if any errors came up
", "rawMarkdown": "Then perhaps the portion of code that saves your submission is not getting run? Check above in your code to see if it all got ran/if any errors came up", "votes": null }, { "id": "906715", "postDate": "06/29/2020 14:02:50", "content": "Ok, thanks.
", "rawMarkdown": "Ok, thanks.", "votes": null }, { "id": "915192", "postDate": "07/04/2020 14:32:18", "content": "Hi, I rewrote with just printing space seperated ones to the submission file. And when I submit this, it gives an error. A zero score is understandable, but why is there an error?
", "rawMarkdown": "Hi, I rewrote with just printing space seperated ones to the submission file. And when I submit this, it gives an error. A zero score is understandable, but why is there an error?", "votes": null }, { "id": "915369", "postDate": "07/04/2020 17:15:54", "content": "A couple reasons that may not be working:\na) Not using the right datatype. The PredictionString should contain a float followed by four integers. If the first number is not a float, it may not work.\nb) Not using the right length. If you tried something like \"0.000 0 0 0 0 0\" it may also not work because there is one extra number.\nc) You are not doing this for the entire test set, only the 10 images that are visible. In order for it to get scored correctly, you have to make predictions for all ~1000 images.\nd) Perhaps since you are passing all zeros, it causes an error during scoring because this indicates you are predicting a box with zero area. It may be a requirement that x0 < x1 and y0 < y1.
\n\nHope this helps.
", "rawMarkdown": "A couple reasons that may not be working:\na) Not using the right datatype. The PredictionString should contain a float followed by four integers. If the first number is not a float, it may not work.\nb) Not using the right length. If you tried something like \"0.000 0 0 0 0 0\" it may also not work because there is one extra number.\nc) You are not doing this for the entire test set, only the 10 images that are visible. In order for it to get scored correctly, you have to make predictions for all ~1000 images.\nd) Perhaps since you are passing all zeros, it causes an error during scoring because this indicates you are predicting a box with zero area. It may be a requirement that x0 < x1 and y0 < y1.\n\nHope this helps.", "votes": null }, { "id": "915455", "postDate": "07/04/2020 18:40:59", "content": "Thanks Matthew for the clear explanations.
", "rawMarkdown": "Thanks Matthew for the clear explanations.", "votes": null }, { "id": "935344", "postDate": "07/19/2020 09:59:52", "content": "Great videos! Thanks.
", "rawMarkdown": "Great videos! Thanks.", "votes": null }, { "id": "958176", "postDate": "08/04/2020 20:20:53", "content": "this thread is really helpful thanks
", "rawMarkdown": "this thread is really helpful thanks", "votes": null }, { "id": "963877", "postDate": "08/09/2020 11:30:19", "content": "Hi,
\nWe want to obtain our fresh score based on the competition metric, so that we can add that in our manuscript.
\nIs it possible to do a fresh submission for public leaderboard (either through Kaggle Notebook or late submission tab)? If so, how can we do it?
\nThanks
\nRonak
Hi,
\nWe want to obtain our fresh score based on the competition metric, so that we can add that in our manuscript.
\nIs it possible to do a fresh submission for public leaderboard (either through Kaggle Notebook or late submission tab)? If so, how can we do it?
\nThanks
\nRonak
Hi, \n I have just started doing basic ML stuff and this is my first time on kaggle.\nI am trying to implement a simple Neural Network. Is the training data given to us or do we have to find it ourselves and then use the add data button in the data section to use it?\nIs the input file empty given to us in the data section empty?\nThanks
", "votes": null, "replies": [ { "id": 834921, "author_name": "rahatreza", "author_url": "", "post_date": "05/05/2020 22:26:34", "content": "We have been given the training data.
", "votes": null, "replies": [] }, { "id": 835066, "author_name": "sushaant", "author_url": "", "post_date": "05/06/2020 03:02:10", "content": "Hi Kanishk,\nsuggest you to go through this.\nKaggle Tutorial - Titanic Competition
", "votes": null, "replies": [] } ] }, { "id": 835096, "author_name": "jagarwal21", "author_url": "", "post_date": "05/06/2020 03:44:12", "content": "Hi ,
\n\nI have just finished my basic ML course and first time in Kaggle. \nI have tried doing Titanic competition and could understand the process. It would be great if you could guide me?
\n\nThanks
", "votes": null, "replies": [ { "id": 835242, "author_name": "bendvd", "author_url": "", "post_date": "05/06/2020 06:19:55", "content": "Happy to have you joining the competition !
\n\nYou can start by looking at these notebooks:\n- https://www.kaggle.com/bendvd/global-wheat-detection-first-submission\n- https://www.kaggle.com/pestipeti/pytorch-starter-fasterrcnn-train
", "votes": null, "replies": [] } ] }, { "id": 837883, "author_name": "ndahiya", "author_url": "", "post_date": "05/08/2020 05:37:51", "content": "I am new to Kaggle and the submission process (or rules??) is not clear at all. The submit to competition button is disabled and mouse hover over it shows: \"You do not have a competition data source.\"\nCan someone explain what that means?
\n\nSimilar to the \"titanic tutorial\" etc linked below I am trying to fine tune a keras_retinanet model and then make predictions. I believe I have successfully created the submissions.csv file.
\n\nAll the tutorials and shared kernels seem to be using Pytorch. But I am trying to use this repo which basically provides pretrained models and framework for finetuning/inference just like Pytorch but with Keras/Tensorflow:
\n\nhttps://github.com/fizyr/keras-retinanet
\n\nI would be happy to share any more details or codes (training/testing anything) etc. if anyone can help me understand what's going on.
\n\nThanks!
", "votes": null, "replies": [] }, { "id": 839011, "author_name": "abhisekjana", "author_url": "", "post_date": "05/09/2020 02:06:25", "content": "When I am turning off internet, the public pre-trained model is not getting downloaded and the script is failing. So that means we can't use transfer learning or there is a workaround ? This works fine with internet on. Please help.
", "votes": null, "replies": [ { "id": 839065, "author_name": "rahatreza", "author_url": "", "post_date": "05/09/2020 03:30:11", "content": "I have seen people using two different notebook instances for training and inference. Internet off criteria is set for inference part only.\nYou can check those notebook on notebooks section.
", "votes": null, "replies": [] }, { "id": 839085, "author_name": "abhisekjana", "author_url": "", "post_date": "05/09/2020 04:01:16", "content": "Thanks for your response.
\n\nSo we can share file between notebooks ? Unfortunately when I saved the version ( run and save ), I couldn't find the weights file in the output folder. I may need to dig in more.
", "votes": null, "replies": [] }, { "id": 839089, "author_name": "matthewmasters", "author_url": "", "post_date": "05/09/2020 04:09:19", "content": "The general strategy here is to have two notebooks: one for training (with internet) and another to perform inference (w/o internet). To get your trained weights to your inference notebook, you should upload them as a Kaggle dataset which your notebook can have access to even with no internet.
\n\nAs Rahat said, there's a few public kernels that use this approach so you should take a look to see how they import the weights.
", "votes": null, "replies": [] }, { "id": 839099, "author_name": "abhisekjana", "author_url": "", "post_date": "05/09/2020 04:25:23", "content": "Appreciate your quick response Matthew. I was able to break the scripts and submit the score.
", "votes": null, "replies": [] } ] }, { "id": 852047, "author_name": "talhakhan1992", "author_url": "", "post_date": "05/18/2020 06:18:15", "content": "Hi\nI am trying to compete in Global Wheat Detection\nIn the sample_submission there are five entries in the predictive strings for each image id.\nI am wondering what are those values?
", "votes": null, "replies": [] }, { "id": 863307, "author_name": "akshatgupta57", "author_url": "", "post_date": "05/27/2020 08:14:00", "content": "Here's my youtube blog companion videos on the competition, helps me be motivated. Might be of help to someone else to get started : https://youtu.be/zxT8_lyA6fc
", "votes": null, "replies": [ { "id": 935344, "author_name": "jonathanrayner", "author_url": "", "post_date": "07/19/2020 09:59:52", "content": "Great videos! Thanks.
", "votes": null, "replies": [] } ] }, { "id": 904964, "author_name": "tejasshyam", "author_url": "", "post_date": "06/28/2020 05:52:13", "content": "I have created my notebook with submissions, but I am unable to find the submit button, I checked the documentation and followed all the rules but I am still unable to submit, can anyone help?
", "votes": null, "replies": [ { "id": 904977, "author_name": "matthewmasters", "author_url": "", "post_date": "06/28/2020 06:26:11", "content": "Commit your notebook. Then go to the viewer and look at the latest version. Scroll down to the output section and click on the 'submission.csv' you generated. Then there should be a submit button that you should see.
", "votes": null, "replies": [] }, { "id": 905438, "author_name": "tejasshyam", "author_url": "", "post_date": "06/28/2020 14:50:01", "content": "I am unable to find this, is it possible for you to share a couple of screenshots with me?\nThis is what my csv looks like:\nhttps://imgur.com/a/QvoUGcd
", "votes": null, "replies": [] }, { "id": 905707, "author_name": "matthewmasters", "author_url": "", "post_date": "06/28/2020 18:20:47", "content": "I think it's because you didn't name it 'submission.csv'. Try again and it should look like this with a blue submit button:\n
Ok, I have tried changing the configuration of my file to \"Save output for this run\" and ranyt it, but I cannot find the output, anything that can be done?
\n\nAlso, sorry to bother you so much with the beginner questions.
", "votes": null, "replies": [] }, { "id": 905771, "author_name": "matthewmasters", "author_url": "", "post_date": "06/28/2020 19:21:05", "content": "Not sure what you mean by the \"Save output for this run\". Did you commit your notebook again?
", "votes": null, "replies": [] }, { "id": 905778, "author_name": "tejasshyam", "author_url": "", "post_date": "06/28/2020 19:31:59", "content": "Yes, I did commit it again with a save for version configuration\nVersion Settings\nSave outputs when creating a Quick Save
\n\nNever save output
\n\nAlways save output
\n\nSave output for this version\nSave & Run All with an accelerator
\n\nWhen I opened advanced settings on the save button, this is what I used to save output for version.
", "votes": null, "replies": [] }, { "id": 905799, "author_name": "matthewmasters", "author_url": "", "post_date": "06/28/2020 20:04:08", "content": "Not really sure what you mean. So when you are editing your notebook, you pressed \"Save Version\" and then \"Save & Run All (Commit)\", right? Then you go to the viewer and look at that last version.
", "votes": null, "replies": [] }, { "id": 906018, "author_name": "tejasshyam", "author_url": "", "post_date": "06/29/2020 03:11:08", "content": "Yes, something like that:\nFirst I opened the advanced settings, and clicked on save output for this version.\nThen I clicked on Save & Run All.
", "votes": null, "replies": [] }, { "id": 906612, "author_name": "matthewmasters", "author_url": "", "post_date": "06/29/2020 12:51:14", "content": "And you're still having trouble finding the submit?
", "votes": null, "replies": [] }, { "id": 906689, "author_name": "tejasshyam", "author_url": "", "post_date": "06/29/2020 13:50:05", "content": "Yeah, I am not able to find the output file.
", "votes": null, "replies": [] }, { "id": 906697, "author_name": "matthewmasters", "author_url": "", "post_date": "06/29/2020 13:53:15", "content": "Does it not generate or what? I'm sorry but I'm not sure how much further I can help. Just try to review what I have already told you
", "votes": null, "replies": [] }, { "id": 906698, "author_name": "tejasshyam", "author_url": "", "post_date": "06/29/2020 13:54:07", "content": "Ok, thanks for your help, Im just confused as to how to save the file once it opens...
", "votes": null, "replies": [] }, { "id": 906701, "author_name": "matthewmasters", "author_url": "", "post_date": "06/29/2020 13:56:06", "content": "Still I'm not sure what you mean.\n- In your notebook, within the code, you should save your predictions as 'submission.csv'.\n- Then commit.\n- Then view your latest commit in the viewer.\n- Scroll down to the output, like you showed me in the screenshot before.\n- Click on the submission.csv file.\n- And click submit.
", "votes": null, "replies": [] }, { "id": 906703, "author_name": "tejasshyam", "author_url": "", "post_date": "06/29/2020 13:57:20", "content": "That was not my submission file, it was the sample submission file that they gave, I am not able to find my submission.csv in the viewer.
", "votes": null, "replies": [] }, { "id": 906712, "author_name": "matthewmasters", "author_url": "", "post_date": "06/29/2020 14:01:13", "content": "Then perhaps the portion of code that saves your submission is not getting run? Check above in your code to see if it all got ran/if any errors came up
", "votes": null, "replies": [] }, { "id": 906715, "author_name": "tejasshyam", "author_url": "", "post_date": "06/29/2020 14:02:50", "content": "Ok, thanks.
", "votes": null, "replies": [] }, { "id": 915192, "author_name": "tejasshyam", "author_url": "", "post_date": "07/04/2020 14:32:18", "content": "Hi, I rewrote with just printing space seperated ones to the submission file. And when I submit this, it gives an error. A zero score is understandable, but why is there an error?
", "votes": null, "replies": [] }, { "id": 915369, "author_name": "matthewmasters", "author_url": "", "post_date": "07/04/2020 17:15:54", "content": "A couple reasons that may not be working:\na) Not using the right datatype. The PredictionString should contain a float followed by four integers. If the first number is not a float, it may not work.\nb) Not using the right length. If you tried something like \"0.000 0 0 0 0 0\" it may also not work because there is one extra number.\nc) You are not doing this for the entire test set, only the 10 images that are visible. In order for it to get scored correctly, you have to make predictions for all ~1000 images.\nd) Perhaps since you are passing all zeros, it causes an error during scoring because this indicates you are predicting a box with zero area. It may be a requirement that x0 < x1 and y0 < y1.
\n\nHope this helps.
", "votes": null, "replies": [] }, { "id": 915455, "author_name": "tejasshyam", "author_url": "", "post_date": "07/04/2020 18:40:59", "content": "Thanks Matthew for the clear explanations.
", "votes": null, "replies": [] } ] }, { "id": 905444, "author_name": "syphax93", "author_url": "", "post_date": "06/28/2020 14:55:18", "content": "Hi everyone
\n\nHere are my two first notebook: \nhttps://www.kaggle.com/syphax93/interactive-visualization-modelling-0-803-top7 \nhttps://www.kaggle.com/syphax93/vapnik-s-theory-bias-variance-and-learning-curve
\n\nI'd like to have your opinion.Thanks
", "votes": null, "replies": [] }, { "id": 958176, "author_name": "utkarshshukla2912", "author_url": "", "post_date": "08/04/2020 20:20:53", "content": "this thread is really helpful thanks
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "833429": "New to machine learning and data science? No question is too basic or too simple. Feel free to start your own thread, or use this thread as a place to post any first-timer clarifying questions for the Kaggle community to help you with! \n\nIf you would consider yourself a beginner but don't know where to get started, let other Kagglers help you take your first steps here!\n\nNew to Kaggle? Take a look at a few videos our very own Dr. Rachael Tatman has put together to learn a bit more about [site etiquette](https://www.youtube.com/watch?v=aIus8si_Et0), [Kaggle lingo](https://www.youtube.com/watch?v=sEJHyuWKd-s), and [how to enter a competition using Kaggle Notebooks](https://www.youtube.com/watch?&v=GJBOMWpLpTQ).", "834790": "Hi, \n I have just started doing basic ML stuff and this is my first time on kaggle.\nI am trying to implement a simple Neural Network. Is the training data given to us or do we have to find it ourselves and then use the add data button in the data section to use it?\nIs the input file empty given to us in the data section empty?\nThanks", "834921": "We have been given the training data.", "835066": "Hi Kanishk,\nsuggest you to go through this.\n[Kaggle Tutorial - Titanic Competition](https://www.kaggle.com/alexisbcook/titanic-tutorial)", "835096": "Hi ,\n\nI have just finished my basic ML course and first time in Kaggle. \nI have tried doing Titanic competition and could understand the process. It would be great if you could guide me?\n\nThanks", "835242": "Happy to have you joining the competition !\n\nYou can start by looking at these notebooks:\n- https://www.kaggle.com/bendvd/global-wheat-detection-first-submission\n- https://www.kaggle.com/pestipeti/pytorch-starter-fasterrcnn-train", "837883": "I am new to Kaggle and the submission process (or rules??) is not clear at all. The submit to competition button is disabled and mouse hover over it shows: \"You do not have a competition data source.\"\nCan someone explain what that means?\n\nSimilar to the \"titanic tutorial\" etc linked below I am trying to fine tune a keras_retinanet model and then make predictions. I believe I have successfully created the submissions.csv file.\n\nAll the tutorials and shared kernels seem to be using Pytorch. But I am trying to use this repo which basically provides pretrained models and framework for finetuning/inference just like Pytorch but with Keras/Tensorflow: \n\nhttps://github.com/fizyr/keras-retinanet\n\nI would be happy to share any more details or codes (training/testing anything) etc. if anyone can help me understand what's going on.\n\nThanks!", "839011": "When I am turning off internet, the public pre-trained model is not getting downloaded and the script is failing. So that means we can't use transfer learning or there is a workaround ? This works fine with internet on. Please help.", "839065": "I have seen people using two different notebook instances for training and inference. Internet off criteria is set for inference part only.\nYou can check those notebook on notebooks section.", "839085": "Thanks for your response.\n\nSo we can share file between notebooks ? Unfortunately when I saved the version ( run and save ), I couldn't find the weights file in the output folder. I may need to dig in more.", "839089": "The general strategy here is to have two notebooks: one for training (with internet) and another to perform inference (w/o internet). To get your trained weights to your inference notebook, you should upload them as a Kaggle dataset which your notebook can have access to even with no internet.\n\nAs Rahat said, there's a few public kernels that use this approach so you should take a look to see how they import the weights.", "839099": "Appreciate your quick response Matthew. I was able to break the scripts and submit the score.", "852047": "Hi\nI am trying to compete in Global Wheat Detection\nIn the sample_submission there are five entries in the predictive strings for each image id.\nI am wondering what are those values?", "863307": "Here's my youtube blog companion videos on the competition, helps me be motivated. Might be of help to someone else to get started : https://youtu.be/zxT8_lyA6fc", "904964": "I have created my notebook with submissions, but I am unable to find the submit button, I checked the documentation and followed all the rules but I am still unable to submit, can anyone help?", "904977": "Commit your notebook. Then go to the viewer and look at the latest version. Scroll down to the output section and click on the 'submission.csv' you generated. Then there should be a submit button that you should see.", "905438": "I am unable to find this, is it possible for you to share a couple of screenshots with me?\nThis is what my csv looks like:\nhttps://imgur.com/a/QvoUGcd", "905444": "Hi everyone \n\nHere are my two first notebook: \nhttps://www.kaggle.com/syphax93/interactive-visualization-modelling-0-803-top7 \nhttps://www.kaggle.com/syphax93/vapnik-s-theory-bias-variance-and-learning-curve \n\nI'd like to have your opinion.Thanks", "905707": "I think it's because you didn't name it 'submission.csv'. Try again and it should look like this with a blue submit button:\n", "905751": "Ok, I have tried changing the configuration of my file to \"Save output for this run\" and ranyt it, but I cannot find the output, anything that can be done?\n\nAlso, sorry to bother you so much with the beginner questions.", "905771": "Not sure what you mean by the \"Save output for this run\". Did you commit your notebook again?", "905778": "Yes, I did commit it again with a save for version configuration\nVersion Settings\nSave outputs when creating a Quick Save\n\n\nNever save output\n\nAlways save output\n\nSave output for this version\nSave & Run All with an accelerator\n\nWhen I opened advanced settings on the save button, this is what I used to save output for version.", "905799": "Not really sure what you mean. So when you are editing your notebook, you pressed \"Save Version\" and then \"Save & Run All (Commit)\", right? Then you go to the viewer and look at that last version.", "906018": "Yes, something like that:\nFirst I opened the advanced settings, and clicked on save output for this version.\nThen I clicked on Save & Run All.", "906612": "And you're still having trouble finding the submit?", "906689": "Yeah, I am not able to find the output file.", "906697": "Does it not generate or what? I'm sorry but I'm not sure how much further I can help. Just try to review what I have already told you", "906698": "Ok, thanks for your help, Im just confused as to how to save the file once it opens...", "906701": "Still I'm not sure what you mean.\n- In your notebook, within the code, you should save your predictions as 'submission.csv'.\n- Then commit.\n- Then view your latest commit in the viewer.\n- Scroll down to the output, like you showed me in the screenshot before.\n- Click on the submission.csv file.\n- And click submit.", "906703": "That was not my submission file, it was the sample submission file that they gave, I am not able to find my submission.csv in the viewer.", "906712": "Then perhaps the portion of code that saves your submission is not getting run? Check above in your code to see if it all got ran/if any errors came up", "906715": "Ok, thanks.", "915192": "Hi, I rewrote with just printing space seperated ones to the submission file. And when I submit this, it gives an error. A zero score is understandable, but why is there an error?", "915369": "A couple reasons that may not be working:\na) Not using the right datatype. The PredictionString should contain a float followed by four integers. If the first number is not a float, it may not work.\nb) Not using the right length. If you tried something like \"0.000 0 0 0 0 0\" it may also not work because there is one extra number.\nc) You are not doing this for the entire test set, only the 10 images that are visible. In order for it to get scored correctly, you have to make predictions for all ~1000 images.\nd) Perhaps since you are passing all zeros, it causes an error during scoring because this indicates you are predicting a box with zero area. It may be a requirement that x0 < x1 and y0 < y1.\n\nHope this helps.", "915455": "Thanks Matthew for the clear explanations.", "935344": "Great videos! Thanks.", "958176": "this thread is really helpful thanks", "963877": "Hi,\nWe want to obtain our fresh score based on the competition metric, so that we can add that in our manuscript.\n\nIs it possible to do a fresh submission for public leaderboard (either through Kaggle Notebook or late submission tab)? If so, how can we do it?\n\nThanks\nRonak" }, "source": "meta" }