__howto.txt

python3 main.py -name MyRun


# serverside runs on cluster:

cd /cluster/home/ruzickav/.keras/models/
wget https://github.com/qubvel/classification_models/releases/download/0.0.1/resnet34_imagenet_1000_no_top.h5

module load python_gpu/3.6.4 hdf5/1.10.1 libpng/1.6.27
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/cluster/apps/python/3.6.4/lib64/
echo $LD_LIBRARY_PATH #to check if it doesn't have mess in it
		(does it have libhdf5.so.101, libpng16.so etc ... ?)
		/cluster/apps/gcc-4.8.5/hdf5-1.10.1-z22q436tpg5j5mrbfk4xex6k2dg73tfo/lib:/cluster/apps/gcc-4.8.5/libpng-1.6.27-d6quyddan2eotsexesxufytusf27xxgz/lib:/cluster/apps/lsf/10.1/linux3.10-glibc2.17-x86_64/lib:/cluster/apps/python/3.6.4/lib64

module load hdf5/1.10.1
module load jpeg/9b

pip3 install git+https://github.com/qubvel/segmentation_models --user
pip3 install opencv-python --user
pip3 install seaborn --user
pip3 install tqdm --user
pip3 install albumentations --user
pip install -U scikit-learn --user
pip install -U scikit-image==0.14.2 --user

bsub (...)

-I is interactive
bsub -n 1 -W 4:00 -R "rusage[mem=32000, ngpus_excl_p=1]" python3 XYZ.py
bsub -I -n 1 -W 4:00 -R "rusage[mem=32000, ngpus_excl_p=1]" python3 main.py -FOLD_I 9 -KFOLDS 10 -model_backend resnet101
bsub -I -n 1 -W 4:00 -R "rusage[mem=32000, ngpus_excl_p=1]" -R "select[gpu_mtotal0>=10240]" python3 main.py -FOLD_I 9 -KFOLDS 10 -model_backend resnet101

bjobs
bbjobs