search

frankkramer-lab / MIScnn

A framework for Medical Image Segmentation with Convolutional Neural Networks and Deep Learning
GNU General Public License v3.0
398 stars 116 forks source link

Kits-19.ipynb #129

Open sreeharih opened 2 years ago

sreeharih commented 2 years ago

Hi, When I am executing the crossvalidation function it is showing the following error. Temporary files are created in the batches directory.

Error_file.txt

muellerdo commented 2 years ago

Short summary from e-mail traffic:

I would assume that the installed Tensorflow version isn't compatible to the Google Colab CUDA version?

Last mail:

I have installed miscnn from the dev branch and changed the tensorflow version to 2.7 but now I am getting the following error.

Try not changing the Tensorflow version in the Colab (so it's compatible with CUDA) and install MIScnn via dev branch.

pip install git+https://github.com/frankkramer-lab/MIScnn.git@development

The dev branch is more "dynamic" to differences in module versions.

Cheers, Dominik

sreeharih commented 2 years ago

Hi, Thanks for the solution the training had started. Now a new problem cropped up I had given k_fold=2, epochs=20, iterations=150, it had completed fold_0 but it didn't start the next fold.It's running for around 16 hours.

Output.txt image

sreeharih commented 2 years ago

Instead of cross_validation () this time I had used model.train() for 20 epochs it's running for 23 hours without any progress.

muellerdo commented 2 years ago

Hello @sreeharih,

any updates on this issue?

I was not able to reproduce this bug of the KiTS19 notebook on a google colab with our dev branch.

Please provide more information on this issue. Is the prepare_subfunction completed and do you have all training samples in your batch directory? Please send a public google colab link for bug reproducibility.

Are you using a free colab account? The KiTS19 is around 200GB and with prepare_subfunction on true, a full copy of the dataset is created. I don't know if Google Colab provide such large disk space for free.

sreeharih commented 2 years ago

Hi Dominik Sorry for the late reply.The problem with the code is still persisting. I am actually using google colab pro version.The prepare_subfunction is completed and all the training samples are created in the batches directory.I tried training the model but it didn't happen .If possible can you please share with me the pretrained model.It will be very helpful for my project work:).

On Wed, Mar 30, 2022 at 2:54 PM Dominik Müller @.***> wrote:

Hello @sreeharih https://github.com/sreeharih,

any updates on this issue?

I was not able to reproduce this bug of the KiTS19 notebook on a google colab with our dev branch.

Please provide more information on this issue. Is the prepare_subfunction completed and do you have all training samples in your batch directory? Please send a public google colab link for bug reproducibility.

Are you using a free colab account? The KiTS19 is around 200GB and with prepare_subfunction on true, a full copy of the dataset is created. I don't know if Google Colab provide such large disk space for free.

— Reply to this email directly, view it on GitHub https://github.com/frankkramer-lab/MIScnn/issues/129#issuecomment-1082834509, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQGRF25RQPQJKH7AXXGBAJDVCQMS7ANCNFSM5QI4LNTQ . You are receiving this because you were mentioned.Message ID: @.***>