![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2034058%2F3675f45c7d823a062c405ecfc24378f2%2Factive.png?generation=1573000190717150&alt=media\")
Stage 2 has officially kicked off. Submissions are now open!
\n\nHere are the details:\n- Dataset Updates:\n - The data page has been updated to include:\n - Stage 2 train set labels CSV: Updated with labels for both Stage 1 train and Stage 1 test. \n - Stage 2 test set images: Sub-folder containing the images you will need to re-run your models on\n - An updated sample submission file for Stage 2.\n - The Stage 2 public test set is intentionally very small. It is only intended to validate that your submission processed without errors. The vast majority of the Stage 2 test set is scored on the private leaderboard, which you will not see until the final submission deadline. \n- How Stage 2 Works:\n - You should have already uploaded and locked in your model(s) at the end of Stage 1. \n - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n - Generate predictions with your model on the Stage 2 test set.\n - Submit your .CSV predictions as you did in Stage 1.\n - Before the Stage 2 final submission deadline (November 11th @ 11:59pm UTC), select your 2 final submissions for scoring on the private leaderboard. Navigate to My Submissions and choose \"Use for final score\" next to the submissions you are selected.\n - At the deadline, the private leaderboard will be revealed.\n - Kaggle completes its standard removal of rules violators, finalizes the leaderboard, and reaches out to winners,\n- Other Important Reminders:\n - You are permitted to retrain your stage 1 models on this updated training set and make adjustments to load the updated train set accordingly. \n - You are not permitted to make any code changes, parameter tuning, hand-labeling, or any other prohibited changes to the model you uploaded at the end of stage 1. \n - Details about how two-stage competitions work are covered extensively in the two-stage FAQ and this helpful summarized forum thread
", "messages": [ { "id": 666258, "postDate": "2019-11-06T00:07:25.010Z", "content": "Stage 2 has officially kicked off. Submissions are now open!
\n\nHere are the details:\n- Dataset Updates:\n - The data page has been updated to include:\n - Stage 2 train set labels CSV: Updated with labels for both Stage 1 train and Stage 1 test. \n - Stage 2 test set images: Sub-folder containing the images you will need to re-run your models on\n - An updated sample submission file for Stage 2.\n - The Stage 2 public test set is intentionally very small. It is only intended to validate that your submission processed without errors. The vast majority of the Stage 2 test set is scored on the private leaderboard, which you will not see until the final submission deadline. \n- How Stage 2 Works:\n - You should have already uploaded and locked in your model(s) at the end of Stage 1. \n - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n - Generate predictions with your model on the Stage 2 test set.\n - Submit your .CSV predictions as you did in Stage 1.\n - Before the Stage 2 final submission deadline (November 11th @ 11:59pm UTC), select your 2 final submissions for scoring on the private leaderboard. Navigate to My Submissions and choose \"Use for final score\" next to the submissions you are selected.\n - At the deadline, the private leaderboard will be revealed.\n - Kaggle completes its standard removal of rules violators, finalizes the leaderboard, and reaches out to winners,\n- Other Important Reminders:\n - You are permitted to retrain your stage 1 models on this updated training set and make adjustments to load the updated train set accordingly. \n - You are not permitted to make any code changes, parameter tuning, hand-labeling, or any other prohibited changes to the model you uploaded at the end of stage 1. \n - Details about how two-stage competitions work are covered extensively in the two-stage FAQ and this helpful summarized forum thread
", "rawMarkdown": "Stage 2 has officially kicked off. Submissions are now open!\n\nHere are the details:\n- **Dataset Updates**:\n - The [data page](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data) has been updated to include:\n - Stage 2 train set labels CSV: Updated with labels for both Stage 1 train and Stage 1 test. \n - Stage 2 test set images: Sub-folder containing the images you will need to re-run your models on\n - An updated sample submission file for Stage 2.\n - The Stage 2 public test set is intentionally very small. It is only intended to validate that your submission processed without errors. The vast majority of the Stage 2 test set is scored on the private leaderboard, which you will not see until the final submission deadline. \n- **How Stage 2 Works**:\n - You should have already uploaded and locked in your model(s) at the end of Stage 1. \n - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n - Generate predictions with your model on the Stage 2 test set.\n - Submit your .CSV predictions as you did in Stage 1.\n - Before the Stage 2 final submission deadline (November 11th @ 11:59pm UTC), select your 2 final submissions for scoring on the private leaderboard. Navigate to [My Submissions](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/submissions) and choose \"Use for final score\" next to the submissions you are selected.\n - At the deadline, the private leaderboard will be revealed.\n - Kaggle completes its standard removal of rules violators, finalizes the leaderboard, and reaches out to winners,\n- **Other Important Reminders**:\n - You are permitted to retrain your stage 1 models on this updated training set and make adjustments to load the updated train set accordingly. \n - You are not permitted to make any code changes, parameter tuning, hand-labeling, or any other prohibited changes to the model you uploaded at the end of stage 1. \n - Details about how two-stage competitions work are covered extensively in the [two-stage FAQ](https://www.kaggle.com/two-stage-frequently-asked-questions) and [this helpful summarized forum thread](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/discussion/115703)", "votes": 17 }, { "id": 666273, "postDate": "2019-11-06T00:31:12.427Z", "content": "Investigating the missing submission button issue.
", "rawMarkdown": "Investigating the missing submission button issue.", "votes": 5, "replies": [ { "id": 666283, "postDate": "2019-11-06T00:41:16.970Z", "content": "@juliaelliott I can see the make submission button
", "rawMarkdown": "@juliaelliott I can see the make submission button" } ] }, { "id": 666270, "postDate": "2019-11-06T00:25:54.637Z", "content": "@juliaelliott I don't have the option to upload a submission.
", "rawMarkdown": "@juliaelliott I don't have the option to upload a submission. ", "votes": 6, "replies": [ { "id": 666271, "postDate": "2019-11-06T00:28:54.450Z", "content": "+1
", "rawMarkdown": "+1", "votes": 3 }, { "id": 666277, "postDate": "2019-11-06T00:32:57.867Z", "content": "+1
", "rawMarkdown": "+1", "votes": 3 }, { "id": 666278, "postDate": "2019-11-06T00:35:44.360Z", "content": "+1
", "rawMarkdown": "+1", "votes": 2 }, { "id": 666280, "postDate": "2019-11-06T00:38:56.820Z", "content": "+1
", "rawMarkdown": "+1", "votes": 2 }, { "id": 666284, "postDate": "2019-11-06T00:42:03.913Z", "content": "Thanks for your feedback, everybody! Should be there now.
", "rawMarkdown": "Thanks for your feedback, everybody! Should be there now.", "votes": 2 }, { "id": 666380, "postDate": "2019-11-06T03:44:07.283Z", "content": "@juliaelliott jul how we know if the team is selected for stg2 .seeing submit pred button will confirm ?
", "rawMarkdown": "@juliaelliott jul how we know if the team is selected for stg2 .seeing submit pred button will confirm ?", "votes": 1 }, { "id": 667150, "postDate": "2019-11-06T21:41:33.800Z", "content": "@jaideepvalani Noone is \"selected\" for Stage 2. If you participated in and uploaded a model at the end of Stage 1, then you are eligible for Stage 2 and should make a submission. If you did not, then you are not eligible and risk being removed from the final leaderboard.
", "rawMarkdown": "@jaideepvalani Noone is \"selected\" for Stage 2. If you participated in and uploaded a model at the end of Stage 1, then you are eligible for Stage 2 and should make a submission. If you did not, then you are not eligible and risk being removed from the final leaderboard." }, { "id": 667767, "postDate": "2019-11-07T17:00:16.237Z", "content": "@juliaelliott i see stage2_submission file is bigger than that of stage1 . Does this sample submission contains private set images id also but score computed only fraction of it which is 1%.
", "rawMarkdown": "@juliaelliott i see stage2_submission file is bigger than that of stage1 . Does this sample submission contains private set images id also but score computed only fraction of it which is 1%. \n" } ] }, { "id": 666920, "postDate": "2019-11-06T16:18:01.187Z", "content": "> - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)
\n\nNo, please no...
\n\nThis doesn't make sense, because anyone with big enough computing power could have just waited for Stage 2, and start working on the competition today. For most of us, the advantage of having worked on Stage 1 could be offset by the ability to retrain on Stage 2 (which I imagine most people will be unable to).
\n\nJust imagine someone with big enough computing power picking up a kernel from Stage 1 and with minimal adaptations retrain it on Stage 2, not only leveraging on the work of others but also beating them by the ability to retrain on a larger dataset in a very short time.
\n\nNo new labels should have been introduced in Stage 2.
", "rawMarkdown": "> - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n\nNo, please no...\n\nThis doesn't make sense, because anyone with big enough computing power could have just waited for Stage 2, and start working on the competition today. For most of us, the advantage of having worked on Stage 1 could be offset by the ability to retrain on Stage 2 (which I imagine most people will be unable to).\n\nJust imagine someone with big enough computing power picking up a kernel from Stage 1 and with minimal adaptations retrain it on Stage 2, not only leveraging on the work of others but also beating them by the ability to retrain on a larger dataset in a very short time.\n\nNo new labels should have been introduced in Stage 2.", "votes": 4, "replies": [ { "id": 666924, "postDate": "2019-11-06T16:19:43.620Z", "content": "I didn't think of this, but you make a shit ton of sense here !
", "rawMarkdown": "I didn't think of this, but you make a shit ton of sense here !\n" }, { "id": 666942, "postDate": "2019-11-06T16:35:55.630Z", "content": "The additional data is only 10%, our initial estimation shows that retraining doesn't help. It probably helps, but only marginally.
Another 10% would come in very handy for validation.
", "rawMarkdown": "Another 10% would come in very handy for validation.", "votes": -1 } ] }, { "id": 666274, "postDate": "2019-11-06T00:31:13.843Z", "content": "i can't see this competition in \"active competition\"\ncan't even submit\nif i expand the leaderboard i see no name,why?\ni uploaded my model+code zip fine in due time(before stage-1 closes)\nplease recheck it @juliaelliott
\n\n
Hi all! Thanks for your patience. The Data page is updated and the new, smaller dataset Julia described is now available.
", "rawMarkdown": "Hi all! Thanks for your patience. The Data page is updated and the new, smaller dataset Julia described is now available.", "votes": 3, "replies": [ { "id": 666745, "postDate": "2019-11-06T13:06:11.667Z", "content": "Hi @philculliton, thanks for the effort. Btw, may I know where is the training data? It seems that they are gone. I could not see it in
\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data
\n\nI also have a problem. I only work with Kaggle kernel and I still could not open a kernel. It says that it is downloading the data, but it takes so much time. Is there any problem? Please see the following
\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/discussion/115964#latest-666643
", "rawMarkdown": "Hi @philculliton, thanks for the effort. Btw, may I know where is the training data? It seems that they are gone. I could not see it in\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data\n\nI also have a problem. I only work with Kaggle kernel and I still could not open a kernel. It says that it is downloading the data, but it takes so much time. Is there any problem? Please see the following\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/discussion/115964#latest-666643", "votes": 3 } ] }, { "id": 666642, "postDate": "2019-11-06T10:11:49.413Z", "content": "When can we expect to have the smaller dataset?
", "rawMarkdown": "When can we expect to have the smaller dataset?", "votes": 3 }, { "id": 666282, "postDate": "2019-11-06T00:40:47.880Z", "content": "Please refresh. \"Submit Predictions\" should now be available. Sorry for the temporary inconvenience.
", "rawMarkdown": "Please refresh. \"Submit Predictions\" should now be available. Sorry for the temporary inconvenience.", "votes": 4, "replies": [ { "id": 666285, "postDate": "2019-11-06T00:42:48.600Z", "content": "It is working, thank you.
", "rawMarkdown": "It is working, thank you." }, { "id": 666289, "postDate": "2019-11-06T00:52:01.880Z", "content": "Me too. Thank you!
", "rawMarkdown": "Me too. Thank you!" } ] }, { "id": 667170, "postDate": "2019-11-06T22:34:06.577Z", "content": "@juliaelliott Hi Julia, we submitted our entire model .hdf5 file for stage 1, but did not submit the code that generated it, is this an issue? We assumed when the directions said to upload your model, it meant to upload the actual model file. Thanks!
", "rawMarkdown": "@juliaelliott Hi Julia, we submitted our entire model .hdf5 file for stage 1, but did not submit the code that generated it, is this an issue? We assumed when the directions said to upload your model, it meant to upload the actual model file. Thanks! ", "votes": 1, "replies": [ { "id": 668688, "postDate": "2019-11-08T18:19:01.620Z", "content": "@philculliton hello, could I please have some clarification on uploading a model vs the code? Thank you
", "rawMarkdown": "@philculliton hello, could I please have some clarification on uploading a model vs the code? Thank you" }, { "id": 668707, "postDate": "2019-11-08T18:52:51.907Z", "content": "@kshannonucsd This will ultimately be the host's call at the end of the competition, if you are in winning standing. They will do the investigation of whether you have made any prohibited code changes between what you uploaded and what you submitted in stage 2. If what you uploaded cannot satisfy that investigation, then you are at risk of disqualification. You may wish to go ahead and make your stage 2 submissions, and see where it lands at the end of the competition.
", "rawMarkdown": "@kshannonucsd This will ultimately be the host's call at the end of the competition, if you are in winning standing. They will do the investigation of whether you have made any prohibited code changes between what you uploaded and what you submitted in stage 2. If what you uploaded cannot satisfy that investigation, then you are at risk of disqualification. You may wish to go ahead and make your stage 2 submissions, and see where it lands at the end of the competition." }, { "id": 668780, "postDate": "2019-11-08T20:56:53.197Z", "content": "Ok thanks for the clarification, I appreciate it.
", "rawMarkdown": "Ok thanks for the clarification, I appreciate it. " } ] }, { "id": 666946, "postDate": "2019-11-06T16:39:39.807Z", "content": "We’ve are aware that some users are experiencing issues with loading the dataset in Notebooks/Kernels. There is a caching issue that we are actively investigating. We will report back when it is resolved. In the meantime, please do not trigger new notebooks/kernels sessions.
", "rawMarkdown": "We’ve are aware that some users are experiencing issues with loading the dataset in Notebooks/Kernels. There is a caching issue that we are actively investigating. We will report back when it is resolved. In the meantime, please do not trigger new notebooks/kernels sessions.", "votes": 1, "replies": [ { "id": 666950, "postDate": "2019-11-06T16:45:03.943Z", "content": "will i get stage-2 train set? i mean full 181 gb stage-2 dataset?
", "rawMarkdown": "will i get stage-2 train set? i mean full 181 gb stage-2 dataset?", "votes": 2 }, { "id": 666959, "postDate": "2019-11-06T16:54:19.457Z", "content": "I do believe that both entire 181Gb as well as test 20Gb will be available soon, otherwise lucky people who got stage 2 train during yesterday and first half of today will crash the LB. After first submissions attempts I'm wondering why LB score is so high (or low like 0.0000) so the re-training attempts are very likelihood to get the best chance for everyone.
", "rawMarkdown": "I do believe that both entire 181Gb as well as test 20Gb will be available soon, otherwise lucky people who got stage 2 train during yesterday and first half of today will crash the LB. After first submissions attempts I'm wondering why LB score is so high (or low like 0.0000) so the re-training attempts are very likelihood to get the best chance for everyone." }, { "id": 666985, "postDate": "2019-11-06T17:28:15.003Z", "content": "we need to download full dataset for local resource to retrain the models as soon as possible, because it takes long time to retrain. otherwise, we can't finish training...
", "rawMarkdown": "we need to download full dataset for local resource to retrain the models as soon as possible, because it takes long time to retrain. otherwise, we can't finish training...", "votes": 1 } ] }, { "id": 666711, "postDate": "2019-11-06T12:02:54.263Z", "content": "I have uploaded my code before deadline in the Team sub-menu. However, I do not see it there. Is it expected?
", "rawMarkdown": "I have uploaded my code before deadline in the Team sub-menu. However, I do not see it there. Is it expected?", "votes": 1 }, { "id": 668107, "postDate": "2019-11-08T01:43:39.233Z", "content": "Hi all!
\n\nPlease let us know if you have questions. We're sorry for the inconvenience - thank you for your patience while we worked on finding solutions.
", "rawMarkdown": "Hi all!\n\n- Full stage 2 train and test datasets are now available, including within notebooks.\n- If you’re looking to access just the stage 2 test images, you should either download the entire dataset, pull each image systematically via notebooks, or you may find it hosted by another fellow participant elsewhere on the forums.\n- Deadline extension remains 11/13 @ 11:59PM UTC\n\nPlease let us know if you have questions. We're sorry for the inconvenience - thank you for your patience while we worked on finding solutions.", "votes": 2, "replies": [ { "id": 668190, "postDate": "2019-11-08T05:20:47.053Z", "content": "@philculliton The Downloading problem is still there. When I switch to GPU instance then a downloading starts where it downloads files which keep on increasing and in that state I can not do anything in the GPU notebook but the time is counted in the weekly quota.\nCan you please look into this problem also.
", "rawMarkdown": "@philculliton The Downloading problem is still there. When I switch to GPU instance then a downloading starts where it downloads files which keep on increasing and in that state I can not do anything in the GPU notebook but the time is counted in the weekly quota.\nCan you please look into this problem also.", "votes": 2 }, { "id": 668197, "postDate": "2019-11-08T05:31:19.817Z", "content": "I have a similar issue. I can see the data available to download, however I cannot access it via notebooks. There is a '? rsna-intracranial-hemorrhage-detection' under the input directory and says \"This type is not yet supported.\"
", "rawMarkdown": "I have a similar issue. I can see the data available to download, however I cannot access it via notebooks. There is a '? rsna-intracranial-hemorrhage-detection' under the input directory and says \"This type is not yet supported.\"", "votes": 2 }, { "id": 668210, "postDate": "2019-11-08T05:53:19.417Z", "content": "@nthomas4 You can read the image data like \"../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train\"\nThis folder contains the stage 2 train dataset.\nI was able to work on these image in the non gpu notebook but when I switch on the GPU, then the weird download thing starts.\nAre you telling about not able to access the data in the GPU instance?
", "rawMarkdown": "@nthomas4 You can read the image data like \"../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train\"\nThis folder contains the stage 2 train dataset.\nI was able to work on these image in the non gpu notebook but when I switch on the GPU, then the weird download thing starts.\nAre you telling about not able to access the data in the GPU instance?" }, { "id": 668217, "postDate": "2019-11-08T06:09:45.747Z", "content": "@saquibhashmi I cannot access in both instances (with/without GPU)
", "rawMarkdown": "@saquibhashmi I cannot access in both instances (with/without GPU) ", "votes": 1 }, { "id": 668229, "postDate": "2019-11-08T06:23:24.160Z", "content": "I think there is still some problem from the kaggle side.\nI hope they better resolve it quickly.
", "rawMarkdown": "I think there is still some problem from the kaggle side.\nI hope they better resolve it quickly." }, { "id": 668466, "postDate": "2019-11-08T13:06:26.223Z", "content": "Hi @nthomas4 and @saquibhashmi - thanks for your feedback, and I'm sorry you're running into trouble.
\n\nWe're looking into your reports on our end. In the meantime @nthomas4, as suggested by @saquibhashmi, have you updated your paths as well? Just to rule out that particular potential issue.
", "rawMarkdown": "Hi @nthomas4 and @saquibhashmi - thanks for your feedback, and I'm sorry you're running into trouble.\n\nWe're looking into your reports on our end. In the meantime @nthomas4, as suggested by @saquibhashmi, have you updated your paths as well? Just to rule out that particular potential issue." }, { "id": 668473, "postDate": "2019-11-08T13:11:50.390Z", "content": "i too face same issue on kaggle GPU.. downloading keeps happening endlessely. I think only those who have got their own Non Kaggle resources are able to submit the scores or futher train the model using test set image data,rest all depending on kaggle gpu are stand still..
", "rawMarkdown": "i too face same issue on kaggle GPU.. downloading keeps happening endlessely. I think only those who have got their own Non Kaggle resources are able to submit the scores or futher train the model using test set image data,rest all depending on kaggle gpu are stand still.." }, { "id": 668527, "postDate": "2019-11-08T14:25:56.107Z", "content": "If people are still experiencing issues (or were and no longer are), please let me know. I tried to fix the downloading issue (you may need to restart a kernel if it's stuck downloading). I'm trying to identify which kernels are still needing to download (most should be getting the dataset instantly).
", "rawMarkdown": "If people are still experiencing issues (or were and no longer are), please let me know. I tried to fix the downloading issue (you may need to restart a kernel if it's stuck downloading). I'm trying to identify which kernels are still needing to download (most should be getting the dataset instantly).", "votes": 1 }, { "id": 668541, "postDate": "2019-11-08T14:43:48.627Z", "content": "Hi @herbison I started my kernel about 3 hours ago...the only issue I encountered was that the path for the input files needed more adjustment then just stage_1 ==> stage_2. But that is a simple fix.
\n\nNo issues with downloading of the data.
", "rawMarkdown": "Hi @herbison I started my kernel about 3 hours ago...the only issue I encountered was that the path for the input files needed more adjustment then just stage_1 ==> stage_2. But that is a simple fix.\n\nNo issues with downloading of the data.", "votes": 2 }, { "id": 668593, "postDate": "2019-11-08T15:46:36.913Z", "content": "@rsmits @herbison Can you please share how you were able to fix the paths in the notebook? I no longer have the endless download issue but still cannot access the data in my kernel.
", "rawMarkdown": "@rsmits @herbison Can you please share how you were able to fix the paths in the notebook? I no longer have the endless download issue but still cannot access the data in my kernel. " }, { "id": 668600, "postDate": "2019-11-08T15:56:32.933Z", "content": "Sorry, just read another post and found the issue. For those that are stuck it is:
\n\n'../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train.csv'
\n\nThanks!
", "rawMarkdown": "Sorry, just read another post and found the issue. For those that are stuck it is:\n \n'../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train.csv'\n\nThanks! ", "votes": 1 }, { "id": 668606, "postDate": "2019-11-08T16:01:00.880Z", "content": "Hi @nthomas4 the root dir was changed to '/kaggle/input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/'.\nAnd some other minor changes in the image directory names.
\n\nMy kernel is still running with the updates...should be finished in little over 4 hours..I hope it finishes succesfully.
", "rawMarkdown": "Hi @nthomas4 the root dir was changed to '/kaggle/input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/'.\nAnd some other minor changes in the image directory names.\n\nMy kernel is still running with the updates...should be finished in little over 4 hours..I hope it finishes succesfully.", "votes": 3 } ] }, { "id": 667026, "postDate": "2019-11-06T18:24:35.723Z", "content": "Hi all,\nI'm wondering whether there are currently issues with the data downloading? I'm unable to download the stage_2_test_images using the api.
\n\nWeirdly enough, this works:\n>kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_sample_submission.csv
However:
\n\nkaggle competitions download -c rsna-intracranial-hemorrhage-detection
and
\n\n>kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images
Both throw 404 Errors.
\n\nThe data page currently also only shows 134 MB of data.
\n\nIs this issue on my side (i.e. should I look into the processes of downloading all the test files one by one) or are there currently some issues?
", "rawMarkdown": "Hi all,\nI'm wondering whether there are currently issues with the data downloading? I'm unable to download the stage_2_test_images using the api. \n\nWeirdly enough, this works:\n>`kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_sample_submission.csv`\n\nHowever:\n\n`kaggle competitions download -c rsna-intracranial-hemorrhage-detection`\n\nand\n\n>`kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images`\n\nBoth throw 404 Errors.\n\nThe data page currently also only shows 134 MB of data. \n\nIs this issue on my side (i.e. should I look into the processes of downloading all the test files one by one) or are there currently some issues?\n\n", "votes": 2, "replies": [ { "id": 667043, "postDate": "2019-11-06T18:46:32.467Z", "content": "I didn't have any trouble downloading the stage 2 data with kaggle competitions download -c rsna-intracranial-hemorrhage-detection (apart from the API Token, for some reason I need to create a new one every time)
I get the same error unfortunately. Can download the csv files without a problem but stage_2_test_images throws 404
", "rawMarkdown": "I get the same error unfortunately. Can download the csv files without a problem but stage\\_2\\_test\\_images throws 404" }, { "id": 667171, "postDate": "2019-11-06T22:36:27.733Z", "content": "Seems to work again! Except for downloading the stag_2_test_images directly (which are anyways the only ones available right now). \nAlso the data page shows 25GB instead of 134MB.
", "rawMarkdown": "Seems to work again! Except for downloading the stag_2_test_images directly (which are anyways the only ones available right now). \nAlso the data page shows 25GB instead of 134MB. " } ] }, { "id": 666889, "postDate": "2019-11-06T15:38:03.643Z", "content": "@juliaelliott Now we can't download stage 2 train dataset? I want to use stage 2 train dataset at GCP built for stage 2.
", "rawMarkdown": "@juliaelliott Now we can't download stage 2 **train** dataset? I want to use stage 2 train dataset at GCP built for stage 2.", "votes": 2 }, { "id": 666266, "postDate": "2019-11-06T00:21:24.450Z", "content": "Thank you for your hard work. I can't see make a submission button on my team's page. Is this still in the works?
", "rawMarkdown": "Thank you for your hard work. I can't see make a submission button on my team's page. Is this still in the works?", "votes": 2 }, { "id": 667643, "postDate": "2019-11-07T13:49:05.973Z", "content": "@juliaelliott @philculliton I noticed that Stage 2 Train data is not available anymore in the Kernels. Are you aware of this as it was available earlier this week? I would like to run my kernel once more on the stage 2 training data.
", "rawMarkdown": "@juliaelliott @philculliton I noticed that Stage 2 Train data is not available anymore in the Kernels. Are you aware of this as it was available earlier this week? I would like to run my kernel once more on the stage 2 training data.", "replies": [ { "id": 667771, "postDate": "2019-11-07T17:08:52.603Z", "content": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline.
", "rawMarkdown": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline." } ] }, { "id": 667403, "postDate": "2019-11-07T07:42:05.210Z", "content": "@juliaelliott I would like to retrain my model on updated Stage 2 train set, as allowed accordingly to the rules for this competition, but training and test data from the first phase are not available anymore. Would you please make them back available again?
", "rawMarkdown": "@juliaelliott I would like to retrain my model on updated Stage 2 train set, as allowed accordingly to the rules for this competition, but training and test data from the first phase are not available anymore. Would you please make them back available again?", "replies": [ { "id": 667770, "postDate": "2019-11-07T17:08:39.063Z", "content": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline.
", "rawMarkdown": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline.", "votes": 1 }, { "id": 667965, "postDate": "2019-11-07T21:08:12.180Z", "content": "Now they are available again, thank you :-)
", "rawMarkdown": "Now they are available again, thank you :-)" } ] }, { "id": 667305, "postDate": "2019-11-07T05:09:51.507Z", "content": "@juliaelliott It looks like that 1% of the stage 2 test data used for public leaderborad is too small, it is possible to observe it based on recent distribution of results. It is not allow to determenistic check correctness of submission if metric is high. Could you consider to increase public test stage 2 dataset from 1% to 5%?
", "rawMarkdown": "@juliaelliott It looks like that 1% of the stage 2 test data used for public leaderborad is too small, it is possible to observe it based on recent distribution of results. It is not allow to determenistic check correctness of submission if metric is high. Could you consider to increase public test stage 2 dataset from 1% to 5%?", "replies": [ { "id": 667775, "postDate": "2019-11-07T17:13:57.113Z", "content": "No sorry, we won’t be increasing the public test set size. It is intentionally very small, purely to ensure your submission can be made without an error. You should not be making any code changes, so it should really be a rerun of your models and submission.
", "rawMarkdown": "No sorry, we won’t be increasing the public test set size. It is intentionally very small, purely to ensure your submission can be made without an error. You should not be making any code changes, so it should really be a rerun of your models and submission. ", "votes": 1 } ] }, { "id": 667271, "postDate": "2019-11-07T03:42:41.493Z", "content": "Do you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?
", "rawMarkdown": "Do you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?", "replies": [ { "id": 667279, "postDate": "2019-11-07T03:56:35.680Z", "content": "\n\n\nClaudio Moisés Valiense de Andrade wrote:
\n \nDo you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?
\n
I see, the stage 2 training files are in the stage 1 test folder
", "rawMarkdown": "> **Claudio Moisés Valiense de Andrade wrote:**\n> \n> Do you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?\n\nI see, the stage 2 training files are in the stage 1 test folder\n" } ] }, { "id": 667233, "postDate": "2019-11-07T01:39:41.673Z", "content": "@juliaelliott I am unable the select submission for my final 2 scores. The checkboxes are disabled.
", "rawMarkdown": "@juliaelliott I am unable the select submission for my final 2 scores. The checkboxes are disabled.", "replies": [ { "id": 667960, "postDate": "2019-11-07T21:04:37.583Z", "content": "Only the team leader can make final submission selections.
", "rawMarkdown": "Only the team leader can make final submission selections." } ] }, { "id": 667058, "postDate": "2019-11-06T19:06:23.670Z", "content": "@juliaelliott I had submitted my model before the deadline to submit the model in Stage 1.\nStill my name is not in the leaderboard of stage 2.\nDo I need to do something else to become eligible to participate in stage 2.\nOr my name will come when I sumit my first submission on stage 2.\nAlso the stage 2 training set is not visible now.
", "rawMarkdown": "@juliaelliott I had submitted my model before the deadline to submit the model in Stage 1.\nStill my name is not in the leaderboard of stage 2.\nDo I need to do something else to become eligible to participate in stage 2.\nOr my name will come when I sumit my first submission on stage 2.\nAlso the stage 2 training set is not visible now.", "replies": [ { "id": 667096, "postDate": "2019-11-06T20:06:11.930Z", "content": "will show after your stage 2 submission.
", "rawMarkdown": "will show after your stage 2 submission." }, { "id": 667155, "postDate": "2019-11-06T21:47:48.287Z", "content": "Ok got it,\nAlso why is the stage 2 training set not visible now.
", "rawMarkdown": "Ok got it,\nAlso why is the stage 2 training set not visible now." } ] }, { "id": 667036, "postDate": "2019-11-06T18:35:43.360Z", "content": "i cant download the stage 2 dataset
", "rawMarkdown": "i cant download the stage 2 dataset" }, { "id": 666738, "postDate": "2019-11-06T12:52:48.923Z", "content": "hi @juliaelliott , As this is my first competition , i don't know that my team is selected for second stage of the competition or not?
", "rawMarkdown": "hi @juliaelliott , As this is my first competition , i don't know that my team is selected for second stage of the competition or not?", "replies": [ { "id": 667147, "postDate": "2019-11-06T21:38:05.117Z", "content": "If you uploaded a model at the end of stage 1, you are eligible and can make a submission to stage 2. If you did not, you are not eligible.
", "rawMarkdown": "If you uploaded a model at the end of stage 1, you are eligible and can make a submission to stage 2. If you did not, you are not eligible." } ] }, { "id": 666424, "postDate": "2019-11-06T04:54:27.800Z", "content": "\n\n\nFor the purposes of medals and points calculations, we will use the number of participants in the first stage to calculate points and medals.
\n
It seems that the medal positions haven't updated on the new public LB yet
", "rawMarkdown": "> For the purposes of medals and points calculations, we will use the number of participants in the first stage to calculate points and medals.\n\nIt seems that the medal positions haven't updated on the new public LB yet", "replies": [ { "id": 666535, "postDate": "2019-11-06T07:35:57.857Z", "content": "It'll be updated later. The usual 2-stage stuff
", "rawMarkdown": "It'll be updated later. The usual 2-stage stuff", "votes": 2 } ] }, { "id": 666311, "postDate": "2019-11-06T01:34:15.470Z", "content": "I still can't get any data yet. Always \"Too many requests\", no matter I use the bottom on the Data page, the API command kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images, or the little python code to request download each image one by one. \nI am starting to worry I did something wrong \"Too many requests\" means I send too many requests in a short time? or just there are too many people trying to download the files in the mean time?\nLooking at people have already submitted their answers and I even can't get any data. It's very frustrating. Especially, the data is large and takes time to retrain the model.
I finally understand what Err 429 means. My bad. I was too anxious to download the data and exceed the limit. I admit I didn't trace back all the discussions in the past two months to find out the proper download command, but tried a snippet of code to request each image one by one, might make the situation worse.\nNow, I found I was rejected to download any competition data from kaggle. So, really need some help here. Does the Err 429 limits goes away in some time? It will be a nightmare if my account was banned due to this mistake.
", "rawMarkdown": "I finally understand what Err 429 means. My bad. I was too anxious to download the data and exceed the limit. I admit I didn't trace back all the discussions in the past two months to find out the proper download command, but tried a snippet of code to request each image one by one, might make the situation worse.\nNow, I found I was rejected to download any competition data from kaggle. So, really need some help here. Does the Err 429 limits goes away in some time? It will be a nightmare if my account was banned due to this mistake. " }, { "id": 667148, "postDate": "2019-11-06T21:38:57.490Z", "content": "@chengjiun Your rate limit should be reset now.
", "rawMarkdown": "@chengjiun Your rate limit should be reset now." } ] }, { "id": 666261, "postDate": "2019-11-06T00:10:04.473Z", "content": "Thanks your hard work!
", "rawMarkdown": "Thanks your hard work!" } ], "comments": [ { "id": 666273, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T00:31:12.427000", "content": "Investigating the missing submission button issue.
", "votes": 5, "replies": [ { "id": 666283, "author_name": "Hilal Shaath", "author_url": "", "post_date": "2019-11-06T00:41:16.970000", "content": "@juliaelliott I can see the make submission button
", "votes": 0, "replies": [] } ] }, { "id": 666270, "author_name": "William Green", "author_url": "", "post_date": "2019-11-06T00:25:54.637000", "content": "@juliaelliott I don't have the option to upload a submission.
", "votes": 6, "replies": [ { "id": 666271, "author_name": "Peter", "author_url": "", "post_date": "2019-11-06T00:28:54.450000", "content": "+1
", "votes": 3, "replies": [] }, { "id": 666277, "author_name": "Hilal Shaath", "author_url": "", "post_date": "2019-11-06T00:32:57.867000", "content": "+1
", "votes": 3, "replies": [] }, { "id": 666278, "author_name": "Kuan-Chieh Huang", "author_url": "", "post_date": "2019-11-06T00:35:44.360000", "content": "+1
", "votes": 2, "replies": [] }, { "id": 666280, "author_name": "Miyabon", "author_url": "", "post_date": "2019-11-06T00:38:56.820000", "content": "+1
", "votes": 2, "replies": [] }, { "id": 666284, "author_name": "Phil Culliton", "author_url": "", "post_date": "2019-11-06T00:42:03.913000", "content": "Thanks for your feedback, everybody! Should be there now.
", "votes": 2, "replies": [] }, { "id": 666380, "author_name": "Jaideep", "author_url": "", "post_date": "2019-11-06T03:44:07.283000", "content": "@juliaelliott jul how we know if the team is selected for stg2 .seeing submit pred button will confirm ?
", "votes": 1, "replies": [] }, { "id": 667150, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T21:41:33.800000", "content": "@jaideepvalani Noone is \"selected\" for Stage 2. If you participated in and uploaded a model at the end of Stage 1, then you are eligible for Stage 2 and should make a submission. If you did not, then you are not eligible and risk being removed from the final leaderboard.
", "votes": 0, "replies": [] }, { "id": 667767, "author_name": "Jaideep", "author_url": "", "post_date": "2019-11-07T17:00:16.237000", "content": "@juliaelliott i see stage2_submission file is bigger than that of stage1 . Does this sample submission contains private set images id also but score computed only fraction of it which is 1%.
", "votes": 0, "replies": [] } ] }, { "id": 666920, "author_name": "Diogo R. Ferreira", "author_url": "", "post_date": "2019-11-06T16:18:01.187000", "content": "> - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)
\n\nNo, please no...
\n\nThis doesn't make sense, because anyone with big enough computing power could have just waited for Stage 2, and start working on the competition today. For most of us, the advantage of having worked on Stage 1 could be offset by the ability to retrain on Stage 2 (which I imagine most people will be unable to).
\n\nJust imagine someone with big enough computing power picking up a kernel from Stage 1 and with minimal adaptations retrain it on Stage 2, not only leveraging on the work of others but also beating them by the ability to retrain on a larger dataset in a very short time.
\n\nNo new labels should have been introduced in Stage 2.
", "votes": 4, "replies": [ { "id": 666924, "author_name": "Benjamin Dubreu", "author_url": "", "post_date": "2019-11-06T16:19:43.620000", "content": "I didn't think of this, but you make a shit ton of sense here !
", "votes": 0, "replies": [] }, { "id": 666942, "author_name": "nosound", "author_url": "", "post_date": "2019-11-06T16:35:55.630000", "content": "The additional data is only 10%, our initial estimation shows that retraining doesn't help. It probably helps, but only marginally.
Another 10% would come in very handy for validation.
", "votes": -1, "replies": [] } ] }, { "id": 666274, "author_name": "Mobassir", "author_url": "", "post_date": "2019-11-06T00:31:13.843000", "content": "i can't see this competition in \"active competition\"\ncan't even submit\nif i expand the leaderboard i see no name,why?\ni uploaded my model+code zip fine in due time(before stage-1 closes)\nplease recheck it @juliaelliott
\n\n
Hi all! Thanks for your patience. The Data page is updated and the new, smaller dataset Julia described is now available.
", "votes": 3, "replies": [ { "id": 666745, "author_name": "arctic", "author_url": "", "post_date": "2019-11-06T13:06:11.667000", "content": "Hi @philculliton, thanks for the effort. Btw, may I know where is the training data? It seems that they are gone. I could not see it in
\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data
\n\nI also have a problem. I only work with Kaggle kernel and I still could not open a kernel. It says that it is downloading the data, but it takes so much time. Is there any problem? Please see the following
\n\nhttps://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/discussion/115964#latest-666643
", "votes": 3, "replies": [] } ] }, { "id": 666642, "author_name": "César Parra Rojas", "author_url": "", "post_date": "2019-11-06T10:11:49.413000", "content": "When can we expect to have the smaller dataset?
", "votes": 3, "replies": [] }, { "id": 666282, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T00:40:47.880000", "content": "Please refresh. \"Submit Predictions\" should now be available. Sorry for the temporary inconvenience.
", "votes": 4, "replies": [ { "id": 666285, "author_name": "Peter", "author_url": "", "post_date": "2019-11-06T00:42:48.600000", "content": "It is working, thank you.
", "votes": 0, "replies": [] }, { "id": 666289, "author_name": "Miyabon", "author_url": "", "post_date": "2019-11-06T00:52:01.880000", "content": "Me too. Thank you!
", "votes": 0, "replies": [] } ] }, { "id": 667170, "author_name": "Kyle Shannon", "author_url": "", "post_date": "2019-11-06T22:34:06.577000", "content": "@juliaelliott Hi Julia, we submitted our entire model .hdf5 file for stage 1, but did not submit the code that generated it, is this an issue? We assumed when the directions said to upload your model, it meant to upload the actual model file. Thanks!
", "votes": 1, "replies": [ { "id": 668688, "author_name": "Kyle Shannon", "author_url": "", "post_date": "2019-11-08T18:19:01.620000", "content": "@philculliton hello, could I please have some clarification on uploading a model vs the code? Thank you
", "votes": 0, "replies": [] }, { "id": 668707, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-08T18:52:51.907000", "content": "@kshannonucsd This will ultimately be the host's call at the end of the competition, if you are in winning standing. They will do the investigation of whether you have made any prohibited code changes between what you uploaded and what you submitted in stage 2. If what you uploaded cannot satisfy that investigation, then you are at risk of disqualification. You may wish to go ahead and make your stage 2 submissions, and see where it lands at the end of the competition.
", "votes": 0, "replies": [] }, { "id": 668780, "author_name": "Kyle Shannon", "author_url": "", "post_date": "2019-11-08T20:56:53.197000", "content": "Ok thanks for the clarification, I appreciate it.
", "votes": 0, "replies": [] } ] }, { "id": 666946, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T16:39:39.807000", "content": "We’ve are aware that some users are experiencing issues with loading the dataset in Notebooks/Kernels. There is a caching issue that we are actively investigating. We will report back when it is resolved. In the meantime, please do not trigger new notebooks/kernels sessions.
", "votes": 1, "replies": [ { "id": 666950, "author_name": "Mobassir", "author_url": "", "post_date": "2019-11-06T16:45:03.943000", "content": "will i get stage-2 train set? i mean full 181 gb stage-2 dataset?
", "votes": 2, "replies": [] }, { "id": 666959, "author_name": "Aleksandr Vystoropskyi", "author_url": "", "post_date": "2019-11-06T16:54:19.457000", "content": "I do believe that both entire 181Gb as well as test 20Gb will be available soon, otherwise lucky people who got stage 2 train during yesterday and first half of today will crash the LB. After first submissions attempts I'm wondering why LB score is so high (or low like 0.0000) so the re-training attempts are very likelihood to get the best chance for everyone.
", "votes": 0, "replies": [] }, { "id": 666985, "author_name": "Inoichan", "author_url": "", "post_date": "2019-11-06T17:28:15.003000", "content": "we need to download full dataset for local resource to retrain the models as soon as possible, because it takes long time to retrain. otherwise, we can't finish training...
", "votes": 1, "replies": [] } ] }, { "id": 666711, "author_name": "EvgenyG", "author_url": "", "post_date": "2019-11-06T12:02:54.263000", "content": "I have uploaded my code before deadline in the Team sub-menu. However, I do not see it there. Is it expected?
", "votes": 1, "replies": [] }, { "id": 668107, "author_name": "Phil Culliton", "author_url": "", "post_date": "2019-11-08T01:43:39.233000", "content": "Hi all!
\n\nPlease let us know if you have questions. We're sorry for the inconvenience - thank you for your patience while we worked on finding solutions.
", "votes": 2, "replies": [ { "id": 668190, "author_name": "Saquib_Hashmi", "author_url": "", "post_date": "2019-11-08T05:20:47.053000", "content": "@philculliton The Downloading problem is still there. When I switch to GPU instance then a downloading starts where it downloads files which keep on increasing and in that state I can not do anything in the GPU notebook but the time is counted in the weekly quota.\nCan you please look into this problem also.
", "votes": 2, "replies": [] }, { "id": 668197, "author_name": "", "author_url": "", "post_date": "2019-11-08T05:31:19.817000", "content": "I have a similar issue. I can see the data available to download, however I cannot access it via notebooks. There is a '? rsna-intracranial-hemorrhage-detection' under the input directory and says \"This type is not yet supported.\"
", "votes": 2, "replies": [] }, { "id": 668210, "author_name": "Saquib_Hashmi", "author_url": "", "post_date": "2019-11-08T05:53:19.417000", "content": "@nthomas4 You can read the image data like \"../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train\"\nThis folder contains the stage 2 train dataset.\nI was able to work on these image in the non gpu notebook but when I switch on the GPU, then the weird download thing starts.\nAre you telling about not able to access the data in the GPU instance?
", "votes": 0, "replies": [] }, { "id": 668217, "author_name": "", "author_url": "", "post_date": "2019-11-08T06:09:45.747000", "content": "@saquibhashmi I cannot access in both instances (with/without GPU)
", "votes": 1, "replies": [] }, { "id": 668229, "author_name": "Saquib_Hashmi", "author_url": "", "post_date": "2019-11-08T06:23:24.160000", "content": "I think there is still some problem from the kaggle side.\nI hope they better resolve it quickly.
", "votes": 0, "replies": [] }, { "id": 668466, "author_name": "Phil Culliton", "author_url": "", "post_date": "2019-11-08T13:06:26.223000", "content": "Hi @nthomas4 and @saquibhashmi - thanks for your feedback, and I'm sorry you're running into trouble.
\n\nWe're looking into your reports on our end. In the meantime @nthomas4, as suggested by @saquibhashmi, have you updated your paths as well? Just to rule out that particular potential issue.
", "votes": 0, "replies": [] }, { "id": 668473, "author_name": "Jaideep", "author_url": "", "post_date": "2019-11-08T13:11:50.390000", "content": "i too face same issue on kaggle GPU.. downloading keeps happening endlessely. I think only those who have got their own Non Kaggle resources are able to submit the scores or futher train the model using test set image data,rest all depending on kaggle gpu are stand still..
", "votes": 0, "replies": [] }, { "id": 668527, "author_name": "Dustin", "author_url": "", "post_date": "2019-11-08T14:25:56.107000", "content": "If people are still experiencing issues (or were and no longer are), please let me know. I tried to fix the downloading issue (you may need to restart a kernel if it's stuck downloading). I'm trying to identify which kernels are still needing to download (most should be getting the dataset instantly).
", "votes": 1, "replies": [] }, { "id": 668541, "author_name": "Robin Smits", "author_url": "", "post_date": "2019-11-08T14:43:48.627000", "content": "Hi @herbison I started my kernel about 3 hours ago...the only issue I encountered was that the path for the input files needed more adjustment then just stage_1 ==> stage_2. But that is a simple fix.
\n\nNo issues with downloading of the data.
", "votes": 2, "replies": [] }, { "id": 668593, "author_name": "", "author_url": "", "post_date": "2019-11-08T15:46:36.913000", "content": "@rsmits @herbison Can you please share how you were able to fix the paths in the notebook? I no longer have the endless download issue but still cannot access the data in my kernel.
", "votes": 0, "replies": [] }, { "id": 668600, "author_name": "", "author_url": "", "post_date": "2019-11-08T15:56:32.933000", "content": "Sorry, just read another post and found the issue. For those that are stuck it is:
\n\n'../input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/stage_2_train.csv'
\n\nThanks!
", "votes": 1, "replies": [] }, { "id": 668606, "author_name": "Robin Smits", "author_url": "", "post_date": "2019-11-08T16:01:00.880000", "content": "Hi @nthomas4 the root dir was changed to '/kaggle/input/rsna-intracranial-hemorrhage-detection/rsna-intracranial-hemorrhage-detection/'.\nAnd some other minor changes in the image directory names.
\n\nMy kernel is still running with the updates...should be finished in little over 4 hours..I hope it finishes succesfully.
", "votes": 3, "replies": [] } ] }, { "id": 667026, "author_name": "srs", "author_url": "", "post_date": "2019-11-06T18:24:35.723000", "content": "Hi all,\nI'm wondering whether there are currently issues with the data downloading? I'm unable to download the stage_2_test_images using the api.
\n\nWeirdly enough, this works:\n>kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_sample_submission.csv
However:
\n\nkaggle competitions download -c rsna-intracranial-hemorrhage-detection
and
\n\n>kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images
Both throw 404 Errors.
\n\nThe data page currently also only shows 134 MB of data.
\n\nIs this issue on my side (i.e. should I look into the processes of downloading all the test files one by one) or are there currently some issues?
", "votes": 2, "replies": [ { "id": 667043, "author_name": "Diogo R. Ferreira", "author_url": "", "post_date": "2019-11-06T18:46:32.467000", "content": "I didn't have any trouble downloading the stage 2 data with kaggle competitions download -c rsna-intracranial-hemorrhage-detection (apart from the API Token, for some reason I need to create a new one every time)
I get the same error unfortunately. Can download the csv files without a problem but stage_2_test_images throws 404
", "votes": 0, "replies": [] }, { "id": 667171, "author_name": "srs", "author_url": "", "post_date": "2019-11-06T22:36:27.733000", "content": "Seems to work again! Except for downloading the stag_2_test_images directly (which are anyways the only ones available right now). \nAlso the data page shows 25GB instead of 134MB.
", "votes": 0, "replies": [] } ] }, { "id": 666889, "author_name": "Inoichan", "author_url": "", "post_date": "2019-11-06T15:38:03.643000", "content": "@juliaelliott Now we can't download stage 2 train dataset? I want to use stage 2 train dataset at GCP built for stage 2.
", "votes": 2, "replies": [] }, { "id": 666266, "author_name": "Hilal Shaath", "author_url": "", "post_date": "2019-11-06T00:21:24.450000", "content": "Thank you for your hard work. I can't see make a submission button on my team's page. Is this still in the works?
", "votes": 2, "replies": [] }, { "id": 667643, "author_name": "Robin Smits", "author_url": "", "post_date": "2019-11-07T13:49:05.973000", "content": "@juliaelliott @philculliton I noticed that Stage 2 Train data is not available anymore in the Kernels. Are you aware of this as it was available earlier this week? I would like to run my kernel once more on the stage 2 training data.
", "votes": 0, "replies": [ { "id": 667771, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-07T17:08:52.603000", "content": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline.
", "votes": 0, "replies": [] } ] }, { "id": 667403, "author_name": "Luca Massaron", "author_url": "", "post_date": "2019-11-07T07:42:05.210000", "content": "@juliaelliott I would like to retrain my model on updated Stage 2 train set, as allowed accordingly to the rules for this competition, but training and test data from the first phase are not available anymore. Would you please make them back available again?
", "votes": 0, "replies": [ { "id": 667770, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-07T17:08:39.063000", "content": "We are working on it, I promise! We will definitely post when it is fully ready. In the meantime, you’ll see we’ve already extended the competition deadline.
", "votes": 1, "replies": [] }, { "id": 667965, "author_name": "Luca Massaron", "author_url": "", "post_date": "2019-11-07T21:08:12.180000", "content": "Now they are available again, thank you :-)
", "votes": 0, "replies": [] } ] }, { "id": 667305, "author_name": "Alexey Kachalov", "author_url": "", "post_date": "2019-11-07T05:09:51.507000", "content": "@juliaelliott It looks like that 1% of the stage 2 test data used for public leaderborad is too small, it is possible to observe it based on recent distribution of results. It is not allow to determenistic check correctness of submission if metric is high. Could you consider to increase public test stage 2 dataset from 1% to 5%?
", "votes": 0, "replies": [ { "id": 667775, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-07T17:13:57.113000", "content": "No sorry, we won’t be increasing the public test set size. It is intentionally very small, purely to ensure your submission can be made without an error. You should not be making any code changes, so it should really be a rerun of your models and submission.
", "votes": 1, "replies": [] } ] }, { "id": 667271, "author_name": "Claudio Moisés Valiense de Andrade", "author_url": "", "post_date": "2019-11-07T03:42:41.493000", "content": "Do you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?
", "votes": 0, "replies": [ { "id": 667279, "author_name": "Claudio Moisés Valiense de Andrade", "author_url": "", "post_date": "2019-11-07T03:56:35.680000", "content": "\n\n\nClaudio Moisés Valiense de Andrade wrote:
\n \nDo you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?
\n
I see, the stage 2 training files are in the stage 1 test folder
", "votes": 0, "replies": [] } ] }, { "id": 667233, "author_name": "Peter", "author_url": "", "post_date": "2019-11-07T01:39:41.673000", "content": "@juliaelliott I am unable the select submission for my final 2 scores. The checkboxes are disabled.
", "votes": 0, "replies": [ { "id": 667960, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-07T21:04:37.583000", "content": "Only the team leader can make final submission selections.
", "votes": 0, "replies": [] } ] }, { "id": 667058, "author_name": "Saquib_Hashmi", "author_url": "", "post_date": "2019-11-06T19:06:23.670000", "content": "@juliaelliott I had submitted my model before the deadline to submit the model in Stage 1.\nStill my name is not in the leaderboard of stage 2.\nDo I need to do something else to become eligible to participate in stage 2.\nOr my name will come when I sumit my first submission on stage 2.\nAlso the stage 2 training set is not visible now.
", "votes": 0, "replies": [ { "id": 667096, "author_name": "catlet", "author_url": "", "post_date": "2019-11-06T20:06:11.930000", "content": "will show after your stage 2 submission.
", "votes": 0, "replies": [] }, { "id": 667155, "author_name": "Saquib_Hashmi", "author_url": "", "post_date": "2019-11-06T21:47:48.287000", "content": "Ok got it,\nAlso why is the stage 2 training set not visible now.
", "votes": 0, "replies": [] } ] }, { "id": 667036, "author_name": "carlos valencia", "author_url": "", "post_date": "2019-11-06T18:35:43.360000", "content": "i cant download the stage 2 dataset
", "votes": 0, "replies": [] }, { "id": 666738, "author_name": "Niklesh Tondare", "author_url": "", "post_date": "2019-11-06T12:52:48.923000", "content": "hi @juliaelliott , As this is my first competition , i don't know that my team is selected for second stage of the competition or not?
", "votes": 0, "replies": [ { "id": 667147, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T21:38:05.117000", "content": "If you uploaded a model at the end of stage 1, you are eligible and can make a submission to stage 2. If you did not, you are not eligible.
", "votes": 0, "replies": [] } ] }, { "id": 666424, "author_name": "datasaurus", "author_url": "", "post_date": "2019-11-06T04:54:27.800000", "content": "\n\n\nFor the purposes of medals and points calculations, we will use the number of participants in the first stage to calculate points and medals.
\n
It seems that the medal positions haven't updated on the new public LB yet
", "votes": 0, "replies": [ { "id": 666535, "author_name": "Stanislav Blinov", "author_url": "", "post_date": "2019-11-06T07:35:57.857000", "content": "It'll be updated later. The usual 2-stage stuff
", "votes": 2, "replies": [] } ] }, { "id": 666311, "author_name": "cjma", "author_url": "", "post_date": "2019-11-06T01:34:15.470000", "content": "I still can't get any data yet. Always \"Too many requests\", no matter I use the bottom on the Data page, the API command kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images, or the little python code to request download each image one by one. \nI am starting to worry I did something wrong \"Too many requests\" means I send too many requests in a short time? or just there are too many people trying to download the files in the mean time?\nLooking at people have already submitted their answers and I even can't get any data. It's very frustrating. Especially, the data is large and takes time to retrain the model.
I finally understand what Err 429 means. My bad. I was too anxious to download the data and exceed the limit. I admit I didn't trace back all the discussions in the past two months to find out the proper download command, but tried a snippet of code to request each image one by one, might make the situation worse.\nNow, I found I was rejected to download any competition data from kaggle. So, really need some help here. Does the Err 429 limits goes away in some time? It will be a nightmare if my account was banned due to this mistake.
", "votes": 0, "replies": [] }, { "id": 667148, "author_name": "Julia Elliott", "author_url": "", "post_date": "2019-11-06T21:38:57.490000", "content": "@chengjiun Your rate limit should be reset now.
", "votes": 0, "replies": [] } ] }, { "id": 666261, "author_name": "Miyabon", "author_url": "", "post_date": "2019-11-06T00:10:04.473000", "content": "Thanks your hard work!
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "666258": "Stage 2 has officially kicked off. Submissions are now open!\n\nHere are the details:\n- **Dataset Updates**:\n - The [data page](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data) has been updated to include:\n - Stage 2 train set labels CSV: Updated with labels for both Stage 1 train and Stage 1 test. \n - Stage 2 test set images: Sub-folder containing the images you will need to re-run your models on\n - An updated sample submission file for Stage 2.\n - The Stage 2 public test set is intentionally very small. It is only intended to validate that your submission processed without errors. The vast majority of the Stage 2 test set is scored on the private leaderboard, which you will not see until the final submission deadline. \n- **How Stage 2 Works**:\n - You should have already uploaded and locked in your model(s) at the end of Stage 1. \n - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n - Generate predictions with your model on the Stage 2 test set.\n - Submit your .CSV predictions as you did in Stage 1.\n - Before the Stage 2 final submission deadline (November 11th @ 11:59pm UTC), select your 2 final submissions for scoring on the private leaderboard. Navigate to [My Submissions](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/submissions) and choose \"Use for final score\" next to the submissions you are selected.\n - At the deadline, the private leaderboard will be revealed.\n - Kaggle completes its standard removal of rules violators, finalizes the leaderboard, and reaches out to winners,\n- **Other Important Reminders**:\n - You are permitted to retrain your stage 1 models on this updated training set and make adjustments to load the updated train set accordingly. \n - You are not permitted to make any code changes, parameter tuning, hand-labeling, or any other prohibited changes to the model you uploaded at the end of stage 1. \n - Details about how two-stage competitions work are covered extensively in the [two-stage FAQ](https://www.kaggle.com/two-stage-frequently-asked-questions) and [this helpful summarized forum thread](https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/discussion/115703)", "666273": "Investigating the missing submission button issue.", "666270": "@juliaelliott I don't have the option to upload a submission. ", "666920": "> - (Optional) Re-train your model on the updated Stage 2 train set (What was the Stage 1 test set is now eligible to include in your training)\n\nNo, please no...\n\nThis doesn't make sense, because anyone with big enough computing power could have just waited for Stage 2, and start working on the competition today. For most of us, the advantage of having worked on Stage 1 could be offset by the ability to retrain on Stage 2 (which I imagine most people will be unable to).\n\nJust imagine someone with big enough computing power picking up a kernel from Stage 1 and with minimal adaptations retrain it on Stage 2, not only leveraging on the work of others but also beating them by the ability to retrain on a larger dataset in a very short time.\n\nNo new labels should have been introduced in Stage 2.", "666274": "i can't see this competition in \"active competition\"\ncan't even submit\nif i expand the leaderboard i see no name,why?\ni uploaded my model+code zip fine in due time(before stage-1 closes)\nplease recheck it @juliaelliott \n\n\n", "666674": "Hi all! Thanks for your patience. The Data page is updated and the new, smaller dataset Julia described is now available.", "666642": "When can we expect to have the smaller dataset?", "666282": "Please refresh. \"Submit Predictions\" should now be available. Sorry for the temporary inconvenience.", "667170": "@juliaelliott Hi Julia, we submitted our entire model .hdf5 file for stage 1, but did not submit the code that generated it, is this an issue? We assumed when the directions said to upload your model, it meant to upload the actual model file. Thanks! ", "666946": "We’ve are aware that some users are experiencing issues with loading the dataset in Notebooks/Kernels. There is a caching issue that we are actively investigating. We will report back when it is resolved. In the meantime, please do not trigger new notebooks/kernels sessions.", "666711": "I have uploaded my code before deadline in the Team sub-menu. However, I do not see it there. Is it expected?", "668107": "Hi all!\n\n- Full stage 2 train and test datasets are now available, including within notebooks.\n- If you’re looking to access just the stage 2 test images, you should either download the entire dataset, pull each image systematically via notebooks, or you may find it hosted by another fellow participant elsewhere on the forums.\n- Deadline extension remains 11/13 @ 11:59PM UTC\n\nPlease let us know if you have questions. We're sorry for the inconvenience - thank you for your patience while we worked on finding solutions.", "667026": "Hi all,\nI'm wondering whether there are currently issues with the data downloading? I'm unable to download the stage_2_test_images using the api. \n\nWeirdly enough, this works:\n>`kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_sample_submission.csv`\n\nHowever:\n\n`kaggle competitions download -c rsna-intracranial-hemorrhage-detection`\n\nand\n\n>`kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images`\n\nBoth throw 404 Errors.\n\nThe data page currently also only shows 134 MB of data. \n\nIs this issue on my side (i.e. should I look into the processes of downloading all the test files one by one) or are there currently some issues?\n\n", "666889": "@juliaelliott Now we can't download stage 2 **train** dataset? I want to use stage 2 train dataset at GCP built for stage 2.", "666266": "Thank you for your hard work. I can't see make a submission button on my team's page. Is this still in the works?", "667643": "@juliaelliott @philculliton I noticed that Stage 2 Train data is not available anymore in the Kernels. Are you aware of this as it was available earlier this week? I would like to run my kernel once more on the stage 2 training data.", "667403": "@juliaelliott I would like to retrain my model on updated Stage 2 train set, as allowed accordingly to the rules for this competition, but training and test data from the first phase are not available anymore. Would you please make them back available again?", "667305": "@juliaelliott It looks like that 1% of the stage 2 test data used for public leaderborad is too small, it is possible to observe it based on recent distribution of results. It is not allow to determenistic check correctness of submission if metric is high. Could you consider to increase public test stage 2 dataset from 1% to 5%?", "667271": "Do you have files in stage 2 that do not exist in the training folder of stage 1, will they put the training data from stage 2?", "667233": "@juliaelliott I am unable the select submission for my final 2 scores. The checkboxes are disabled.", "667058": "@juliaelliott I had submitted my model before the deadline to submit the model in Stage 1.\nStill my name is not in the leaderboard of stage 2.\nDo I need to do something else to become eligible to participate in stage 2.\nOr my name will come when I sumit my first submission on stage 2.\nAlso the stage 2 training set is not visible now.", "667036": "i cant download the stage 2 dataset", "666738": "hi @juliaelliott , As this is my first competition , i don't know that my team is selected for second stage of the competition or not?", "666424": "> For the purposes of medals and points calculations, we will use the number of participants in the first stage to calculate points and medals.\n\nIt seems that the medal positions haven't updated on the new public LB yet", "666311": "I still can't get any data yet. Always \"Too many requests\", no matter I use the bottom on the Data page, the API command `kaggle competitions download -c rsna-intracranial-hemorrhage-detection -f stage_2_test_images`, or the little python code to request download each image one by one. \nI am starting to worry I did something wrong \"Too many requests\" means I send too many requests in a short time? or just there are too many people trying to download the files in the mean time?\nLooking at people have already submitted their answers and I even can't get any data. It's very frustrating. Especially, the data is large and takes time to retrain the model.", "666261": "Thanks your hard work!" } }