\n
![\"sc\"](\"https://i.postimg.cc/2yNZM53D/Screen-Shot-2021-03-21-at-10-10-27.png\")
Does anyone succeed in submission recently? I resubmitted my old notebooks and other public notebooks, but can not get any score. Is there something wrong or I missed something?
\n
Does anyone succeed in submission recently? I resubmitted my old notebooks and other public notebooks, but can not get any score. Is there something wrong or I missed something?
\n
I keep getting \"Notebook Exceeded Allowed Compute\" even though I use models and data sizes that are smaller than my previously submitted notebooks. Not sure what to do. I store the image batches as tf records and delete each file after prediction. I use rle_encode_less_memory().
", "rawMarkdown": "I keep getting \"Notebook Exceeded Allowed Compute\" even though I use models and data sizes that are smaller than my previously submitted notebooks. Not sure what to do. I store the image batches as tf records and delete each file after prediction. I use rle_encode_less_memory().", "votes": null }, { "id": "1246695", "postDate": "03/21/2021 02:32:24", "content": "Same thing here. I tried with exact same code that was success before and it cannot stop running.
", "rawMarkdown": "Same thing here. I tried with exact same code that was success before and it cannot stop running.", "votes": null }, { "id": "1246700", "postDate": "03/21/2021 02:39:37", "content": "Using trycatch, my notebook could finish, meaning that some errors make the notebook \"keep running\" indefinitely, but the errors are there. Probably kaggles side problem. For example, imread from tifffile works with all 15 datasets from train data. But on private, either because of some configurations, or missing dependences, it doesnt run. ( for me, that are using R and the package reticulate to run python chunks on R)
", "rawMarkdown": "Using trycatch, my notebook could finish, meaning that some errors make the notebook \"keep running\" indefinitely, but the errors are there. Probably kaggles side problem. For example, imread from tifffile works with all 15 datasets from train data. But on private, either because of some configurations, or missing dependences, it doesnt run. ( for me, that are using R and the package reticulate to run python chunks on R)", "votes": null }, { "id": "1246714", "postDate": "03/21/2021 02:58:39", "content": "In the morning today (Saturday 3/20 in US), I could run&save a couple of codes, but their submission show error and could not finish. In the afternoon, my previously-run codes could not even finish the run&save.
", "rawMarkdown": "In the morning today (Saturday 3/20 in US), I could run&save a couple of codes, but their submission show error and could not finish. In the afternoon, my previously-run codes could not even finish the run&save.", "votes": null }, { "id": "1249974", "postDate": "03/23/2021 17:05:46", "content": "How did you solve that? Im using tifffile imread too and my submission is not working.
", "rawMarkdown": "How did you solve that? Im using tifffile imread too and my submission is not working.", "votes": null }, { "id": "1250222", "postDate": "03/23/2021 21:01:17", "content": "Trycatch in the exact same code. If you are using R, u can try a wrap function to trycatch, purrr::possibly, which is easier. Also, for both R or python, a trycatch routine can be found over the internet to do what u need
", "rawMarkdown": "Trycatch in the exact same code. If you are using R, u can try a wrap function to trycatch, purrr::possibly, which is easier. Also, for both R or python, a trycatch routine can be found over the internet to do what u need", "votes": null }, { "id": "1250226", "postDate": "03/23/2021 21:05:53", "content": "Im sorry, i didnt get it. Im using try catch and i know that the problem is when i read the file in private test set. I only dont know how do i solve that. Do i need to jump for the next image? Doing that doesnt make the score drop abruptly?
", "rawMarkdown": "Im sorry, i didnt get it. Im using try catch and i know that the problem is when i read the file in private test set. I only dont know how do i solve that. Do i need to jump for the next image? Doing that doesnt make the score drop abruptly?", "votes": null }, { "id": "1251158", "postDate": "03/24/2021 14:29:00", "content": "Im not sure if the problem is when reading. It can possibly be:
\n1- installing the new packages from datasets ( can generate some warning)
\n1- reading tifffile ( dont believe that it is size problem here)
\n2- transforming to a tensorflow object (if in R), since got a warning of integer overflow
\n3- some bug that is blocking the notebook to finish when founding any of this above.
\nSo, i made a function that do all those steps, generating the rle predicitons for each image and manage the errors/warnings with this output .
I keep getting \"Notebook Exceeded Allowed Compute\" even though I use models and data sizes that are smaller than my previously submitted notebooks. Not sure what to do. I store the image batches as tf records and delete each file after prediction. I use rle_encode_less_memory().
", "votes": null, "replies": [] }, { "id": 1246695, "author_name": "reaverlee", "author_url": "", "post_date": "03/21/2021 02:32:24", "content": "Same thing here. I tried with exact same code that was success before and it cannot stop running.
", "votes": null, "replies": [] }, { "id": 1246700, "author_name": "rpsantosakaggle", "author_url": "", "post_date": "03/21/2021 02:39:37", "content": "Using trycatch, my notebook could finish, meaning that some errors make the notebook \"keep running\" indefinitely, but the errors are there. Probably kaggles side problem. For example, imread from tifffile works with all 15 datasets from train data. But on private, either because of some configurations, or missing dependences, it doesnt run. ( for me, that are using R and the package reticulate to run python chunks on R)
", "votes": null, "replies": [ { "id": 1249974, "author_name": "joaorodriguez", "author_url": "", "post_date": "03/23/2021 17:05:46", "content": "How did you solve that? Im using tifffile imread too and my submission is not working.
", "votes": null, "replies": [] }, { "id": 1250222, "author_name": "rpsantosakaggle", "author_url": "", "post_date": "03/23/2021 21:01:17", "content": "Trycatch in the exact same code. If you are using R, u can try a wrap function to trycatch, purrr::possibly, which is easier. Also, for both R or python, a trycatch routine can be found over the internet to do what u need
", "votes": null, "replies": [] }, { "id": 1250226, "author_name": "joaorodriguez", "author_url": "", "post_date": "03/23/2021 21:05:53", "content": "Im sorry, i didnt get it. Im using try catch and i know that the problem is when i read the file in private test set. I only dont know how do i solve that. Do i need to jump for the next image? Doing that doesnt make the score drop abruptly?
", "votes": null, "replies": [] }, { "id": 1251158, "author_name": "rpsantosakaggle", "author_url": "", "post_date": "03/24/2021 14:29:00", "content": "Im not sure if the problem is when reading. It can possibly be:
\n1- installing the new packages from datasets ( can generate some warning)
\n1- reading tifffile ( dont believe that it is size problem here)
\n2- transforming to a tensorflow object (if in R), since got a warning of integer overflow
\n3- some bug that is blocking the notebook to finish when founding any of this above.
\nSo, i made a function that do all those steps, generating the rle predicitons for each image and manage the errors/warnings with this output .
In the morning today (Saturday 3/20 in US), I could run&save a couple of codes, but their submission show error and could not finish. In the afternoon, my previously-run codes could not even finish the run&save.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1246660": "Does anyone succeed in submission recently? I resubmitted my old notebooks and other public notebooks, but can not get any score. Is there something wrong or I missed something? \n", "1246671": "I keep getting \"Notebook Exceeded Allowed Compute\" even though I use models and data sizes that are smaller than my previously submitted notebooks. Not sure what to do. I store the image batches as tf records and delete each file after prediction. I use rle_encode_less_memory().", "1246695": "Same thing here. I tried with exact same code that was success before and it cannot stop running.", "1246700": "Using trycatch, my notebook could finish, meaning that some errors make the notebook \"keep running\" indefinitely, but the errors are there. Probably kaggles side problem. For example, imread from tifffile works with all 15 datasets from train data. But on private, either because of some configurations, or missing dependences, it doesnt run. ( for me, that are using R and the package reticulate to run python chunks on R)", "1246714": "In the morning today (Saturday 3/20 in US), I could run&save a couple of codes, but their submission show error and could not finish. In the afternoon, my previously-run codes could not even finish the run&save.", "1249974": "How did you solve that? Im using tifffile imread too and my submission is not working.", "1250222": "Trycatch in the exact same code. If you are using R, u can try a wrap function to trycatch, purrr::possibly, which is easier. Also, for both R or python, a trycatch routine can be found over the internet to do what u need", "1250226": "Im sorry, i didnt get it. Im using try catch and i know that the problem is when i read the file in private test set. I only dont know how do i solve that. Do i need to jump for the next image? Doing that doesnt make the score drop abruptly?", "1251158": "Im not sure if the problem is when reading. It can possibly be:\n1- installing the new packages from datasets ( can generate some warning)\n1- reading tifffile ( dont believe that it is size problem here)\n2- transforming to a tensorflow object (if in R), since got a warning of integer overflow\n3- some bug that is blocking the notebook to finish when founding any of this above.\nSo, i made a function that do all those steps, generating the rle predicitons for each image and manage the errors/warnings with this output ." }, "source": "meta" }