I have shared my basic approach to getting started in these competitions on Google Colab here: https://www.kaggle.com/reubenschmidt/getting-started-in-rsna-miccai-with-google-colab There are many advantages to this approach, including download speeds of ~100Mb/s.
", "messages": [ { "id": "1387384", "postDate": "07/14/2021 06:28:53", "content": "I have shared my basic approach to getting started in these competitions on Google Colab here: https://www.kaggle.com/reubenschmidt/getting-started-in-rsna-miccai-with-google-colab There are many advantages to this approach, including download speeds of ~100Mb/s.
", "rawMarkdown": "I have shared my basic approach to getting started in these competitions on Google Colab here: https://www.kaggle.com/reubenschmidt/getting-started-in-rsna-miccai-with-google-colab There are many advantages to this approach, including download speeds of ~100Mb/s.", "votes": null }, { "id": "1388129", "postDate": "07/14/2021 17:01:13", "content": "Awesome points @reubenschmidt! I love the free GPU they give us. It makes training computer vision model much easier.
", "rawMarkdown": "Awesome points @reubenschmidt! I love the free GPU they give us. It makes training computer vision model much easier.", "votes": null }, { "id": "1507656", "postDate": "09/09/2021 11:45:51", "content": "Hi @reubenschmidt have tried some months ago run one kaggle project in collab, the thing is that there is no much problem until you'd try to train your model.
\nThe main problem is that google detect that the storage that you are using belongs to them (File system scheme '[local]' not implemented), and they invited to you to pay for an google cloud storage account. In theory, there are some free slots if your storage is not so big, but also you have to put your credit card number. I also have found some work around (https://stackoverflow.com/questions/53003208/how-to-save-a-tensorflow-checkpoint-file-from-google-colaboratory-in-when-using) to this problems related with TPU url, but this solutions have some years old, I guess that google people path this thing, because I couldn't work my train model.
", "rawMarkdown": "Hi @reubenschmidt have tried some months ago run one kaggle project in collab, the thing is that there is no much problem until you'd try to train your model.\n\nThe main problem is that google detect that the storage that you are using belongs to them (File system scheme '[local]' not implemented), and they invited to you to pay for an google cloud storage account. In theory, there are some free slots if your storage is not so big, but also you have to put your credit card number. I also have found some work around (https://stackoverflow.com/questions/53003208/how-to-save-a-tensorflow-checkpoint-file-from-google-colaboratory-in-when-using) to this problems related with TPU url, but this solutions have some years old, I guess that google people path this thing, because I couldn't work my train model.", "votes": null } ], "comments": [ { "id": 1388129, "author_name": "jonaslneri", "author_url": "", "post_date": "07/14/2021 17:01:13", "content": "Awesome points @reubenschmidt! I love the free GPU they give us. It makes training computer vision model much easier.
", "votes": null, "replies": [] }, { "id": 1507656, "author_name": "victorfernandezalbor", "author_url": "", "post_date": "09/09/2021 11:45:51", "content": "Hi @reubenschmidt have tried some months ago run one kaggle project in collab, the thing is that there is no much problem until you'd try to train your model.
\nThe main problem is that google detect that the storage that you are using belongs to them (File system scheme '[local]' not implemented), and they invited to you to pay for an google cloud storage account. In theory, there are some free slots if your storage is not so big, but also you have to put your credit card number. I also have found some work around (https://stackoverflow.com/questions/53003208/how-to-save-a-tensorflow-checkpoint-file-from-google-colaboratory-in-when-using) to this problems related with TPU url, but this solutions have some years old, I guess that google people path this thing, because I couldn't work my train model.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1387384": "I have shared my basic approach to getting started in these competitions on Google Colab here: https://www.kaggle.com/reubenschmidt/getting-started-in-rsna-miccai-with-google-colab There are many advantages to this approach, including download speeds of ~100Mb/s.", "1388129": "Awesome points @reubenschmidt! I love the free GPU they give us. It makes training computer vision model much easier.", "1507656": "Hi @reubenschmidt have tried some months ago run one kaggle project in collab, the thing is that there is no much problem until you'd try to train your model.\n\nThe main problem is that google detect that the storage that you are using belongs to them (File system scheme '[local]' not implemented), and they invited to you to pay for an google cloud storage account. In theory, there are some free slots if your storage is not so big, but also you have to put your credit card number. I also have found some work around (https://stackoverflow.com/questions/53003208/how-to-save-a-tensorflow-checkpoint-file-from-google-colaboratory-in-when-using) to this problems related with TPU url, but this solutions have some years old, I guess that google people path this thing, because I couldn't work my train model." }, "source": "meta" }