Hihi, I am facing problem to submit kernel,\nafter I clicked the commit button and waited for a while\nand scrolled to output then tried to click submit to competition,\nbut the text color of these words are gray and cannot click.\nIn addition, I moved the mouse on the button and got a message \n\"your kernel cannot use internet access for this competition\".
\n\nCould anyone give me any suggestion?\nThank you all !!
", "messages": [ { "id": "574229", "postDate": "07/13/2019 13:31:02", "content": "Hihi, I am facing problem to submit kernel,\nafter I clicked the commit button and waited for a while\nand scrolled to output then tried to click submit to competition,\nbut the text color of these words are gray and cannot click.\nIn addition, I moved the mouse on the button and got a message \n\"your kernel cannot use internet access for this competition\".
\n\nCould anyone give me any suggestion?\nThank you all !!
", "rawMarkdown": "Hihi, I am facing problem to submit kernel,\nafter I clicked the commit button and waited for a while\nand scrolled to output then tried to click submit to competition,\nbut the text color of these words are gray and cannot click.\nIn addition, I moved the mouse on the button and got a message \n\"your kernel cannot use internet access for this competition\".\n\nCould anyone give me any suggestion?\nThank you all !!", "votes": null }, { "id": "574614", "postDate": "07/14/2019 07:15:28", "content": "I have the same problem!Did you solved it?
", "rawMarkdown": "I have the same problem!Did you solved it?", "votes": null }, { "id": "574667", "postDate": "07/14/2019 09:21:52", "content": "Internet is off for this Kernel Only competition.\nTrain your model offline or in separate kernel and upload the model weights as dataset.\nBuild kernel for inference using the uploaded model and submit without internet access.
", "rawMarkdown": "Internet is off for this Kernel Only competition.\nTrain your model offline or in separate kernel and upload the model weights as dataset.\nBuild kernel for inference using the uploaded model and submit without internet access.", "votes": null }, { "id": "583800", "postDate": "07/25/2019 02:15:22", "content": "Is there a tutorial/document/sample kernel I am a beginner, and now stuck in saving the model and loading it back?
\n\nI tried torch.save() and torch.load() but the saved model is lost after a kernel restart.
Hey @shankarps I've been there. The files which your kernel saves are temporary, I remember there's a thread where people have discussed how to download files from Kaggle kernels. (Just go through the discussion forums and 99.99% of such problems would be solved easily. )
", "rawMarkdown": "Hey @shankarps I've been there. The files which your kernel saves are temporary, I remember there's a thread where people have discussed how to download files from Kaggle kernels. (Just go through the discussion forums and 99.99% of such problems would be solved easily. )", "votes": null }, { "id": "584365", "postDate": "07/25/2019 20:25:49", "content": "Thanks for the tip @rishabhiitbhu I figured it out. For anyone looking to save and load a trained model, please see the following steps I used.
\n\ntorch.save().Refer to these kernels for sample code -
\n\nTrain and save model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-train
\n\nLoad and use model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-test-phase
", "rawMarkdown": "Thanks for the tip @rishabhiitbhu I figured it out. For anyone looking to save and load a trained model, please see the following steps I used.\n\n1. Create a kernel with internet enabled, initialize your model and train it.\n2. Save the model with `torch.save()`.\n3. Commit and run the kernel. **This is important. Just running the kernel on the screen and saving is not enough. The file will be lost.**\n4. After the commit runs, go to the output files of the commit and download the saved model file.\n5. Create a new kernel with internet off.\n6. Upload the model by clicking \"Add Dataset\" \n6. Create the same model type as empty model with pretrained = False.\n7. Load the saved, uploaded model as torch.load()\n\nRefer to these kernels for sample code - \n\nTrain and save model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-train\n\nLoad and use model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-test-phase", "votes": null }, { "id": "584517", "postDate": "07/26/2019 05:37:17", "content": "You can download model file without committing. Check this out: https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/99946
", "rawMarkdown": "You can download model file without committing. Check this out: https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/99946", "votes": null } ], "comments": [ { "id": 574614, "author_name": "bcwang", "author_url": "", "post_date": "07/14/2019 07:15:28", "content": "I have the same problem!Did you solved it?
", "votes": null, "replies": [] }, { "id": 574667, "author_name": "mktripathi456", "author_url": "", "post_date": "07/14/2019 09:21:52", "content": "Internet is off for this Kernel Only competition.\nTrain your model offline or in separate kernel and upload the model weights as dataset.\nBuild kernel for inference using the uploaded model and submit without internet access.
", "votes": null, "replies": [ { "id": 583800, "author_name": "shankarps", "author_url": "", "post_date": "07/25/2019 02:15:22", "content": "Is there a tutorial/document/sample kernel I am a beginner, and now stuck in saving the model and loading it back?
\n\nI tried torch.save() and torch.load() but the saved model is lost after a kernel restart.
Hey @shankarps I've been there. The files which your kernel saves are temporary, I remember there's a thread where people have discussed how to download files from Kaggle kernels. (Just go through the discussion forums and 99.99% of such problems would be solved easily. )
", "votes": null, "replies": [] }, { "id": 584365, "author_name": "shankarps", "author_url": "", "post_date": "07/25/2019 20:25:49", "content": "Thanks for the tip @rishabhiitbhu I figured it out. For anyone looking to save and load a trained model, please see the following steps I used.
\n\ntorch.save().Refer to these kernels for sample code -
\n\nTrain and save model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-train
\n\nLoad and use model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-test-phase
", "votes": null, "replies": [] }, { "id": 584517, "author_name": "rishabhiitbhu", "author_url": "", "post_date": "07/26/2019 05:37:17", "content": "You can download model file without committing. Check this out: https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/99946
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "574229": "Hihi, I am facing problem to submit kernel,\nafter I clicked the commit button and waited for a while\nand scrolled to output then tried to click submit to competition,\nbut the text color of these words are gray and cannot click.\nIn addition, I moved the mouse on the button and got a message \n\"your kernel cannot use internet access for this competition\".\n\nCould anyone give me any suggestion?\nThank you all !!", "574614": "I have the same problem!Did you solved it?", "574667": "Internet is off for this Kernel Only competition.\nTrain your model offline or in separate kernel and upload the model weights as dataset.\nBuild kernel for inference using the uploaded model and submit without internet access.", "583800": "Is there a tutorial/document/sample kernel I am a beginner, and now stuck in saving the model and loading it back?\n\nI tried `torch.save()` and `torch.load()` but the saved model is lost after a kernel restart.", "583887": "Hey @shankarps I've been there. The files which your kernel saves are temporary, I remember there's a thread where people have discussed how to download files from Kaggle kernels. (Just go through the discussion forums and 99.99% of such problems would be solved easily. )", "584365": "Thanks for the tip @rishabhiitbhu I figured it out. For anyone looking to save and load a trained model, please see the following steps I used.\n\n1. Create a kernel with internet enabled, initialize your model and train it.\n2. Save the model with `torch.save()`.\n3. Commit and run the kernel. **This is important. Just running the kernel on the screen and saving is not enough. The file will be lost.**\n4. After the commit runs, go to the output files of the commit and download the saved model file.\n5. Create a new kernel with internet off.\n6. Upload the model by clicking \"Add Dataset\" \n6. Create the same model type as empty model with pretrained = False.\n7. Load the saved, uploaded model as torch.load()\n\nRefer to these kernels for sample code - \n\nTrain and save model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-train\n\nLoad and use model - https://www.kaggle.com/shankarps/aptos-2019-begginer-pytorch-kernel-test-phase", "584517": "You can download model file without committing. Check this out: https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/99946" }, "source": "meta" }