luminoth
luminoth copied to clipboard
tryolabs
Luminoth does not recognize tensorflow-gpu installation
I have installed tensorflow-gpu (1.12.0) and luminoth (0.2.3) via pip. I get the following message:
Luminoth requires a TensorFlow >= 1.5 installation.
Depending on your use case, you should install either
tensorflowortensorflow-gpupackages manually or via PyPI.
How do I get luminoth to recognize the GPU version of tensorflow?
When I install just tensorflow, luminoth does recognize the package. When both are installed, it does not seem to run on tensorflow-gpu however.
I had the same problem, you do not have CUDA drivers installed correctly. Then works fine with tensrorflow-gpu even less than 1.5
I had the same problem, you do not have CUDA drivers installed correctly. Then works fine with tensrorflow-gpu even less than 1.5
Can you explain how to install it correctly please. I presume to just follow what is on the site where you download cuda.
I solved this following tensorflow installation steps in https://www.tensorflow.org/install
I installed twnsorflow using pip and it works.
I solved this with reinstalling tensorflow-gpu, installing CUDA Drivers and cuDNN. I missed CUDA and cuDNN libraries for the driver. Only driver installation won't work for tensorflow.
Refer: https://www.tensorflow.org/install/gpu
Windows 10 Nvidia GeForce 610 card CUDA 10.0 (Latest) cuDNN 7.5 for Driver version 10 (Windows) Tensorflow 1.13.0 (Current latest) Luminoth 0.2.3
Anyone get a definitive fix for this? I'm having the same problem. Have tried re-installing all the drivers and tensorflow-gpu and still doesn't work. On Ubuntu 18.04 w/ 1080 Ti
Just to make sure, are you installing it in any conda environments you created? I had a similar issue while installing luminoth and tensorflow in a conda environment. Deactivating the environment and reactivating it solved the problem.
If you are running via a jupyter notebook, try checking if luminoth has been installed. You can do this in the runtime by running pip list. If tensorflow or tensorflow-gpu doesn't show up in the list, install it again in the runtime and try.
I'm running into this issue as well.
Environment details:
AWS Deep Learning Base AMI (Amazon Linux) Version 19.0 (ami-0bdd095c2fbd0e692)
$ pip freeze
absl-py==0.8.0
astor==0.8.0
certifi==2019.6.16
chardet==3.0.4
click==6.7
dm-sonnet==1.23
easydict==1.8
Flask==1.1.1
gast==0.2.2
google-pasta==0.1.7
grpcio==1.23.0
h5py==2.9.0
idna==2.8
itsdangerous==1.1.0
Jinja2==2.10.1
Keras-Applications==1.0.8
Keras-Preprocessing==1.1.0
luminoth==0.2.3
lxml==4.4.1
Markdown==3.1.1
MarkupSafe==1.1.1
numpy==1.17.1
Pillow==6.1.0
protobuf==3.9.1
PyYAML==3.13
requests==2.22.0
scikit-video==1.1.11
scipy==1.3.1
six==1.12.0
tensorboard==1.14.0
tensorflow-estimator==1.14.0
tensorflow-gpu==1.14.0
termcolor==1.1.0
urllib3==1.25.3
Werkzeug==0.15.5
wrapt==1.11.2
$ python --version
Python 3.7.4
When I run $ lumi I get the following error:
Luminoth requires a TensorFlow >= 1.5 installation.
Depending on your use case, you should install either `tensorflow` or
`tensorflow-gpu` packages manually or via PyPI.
I've already tried uninstalling and reinstalling tensorflow-gpu, but I keep getting the same error. I also tried uninstalling tensorflow-gpu and installing tensorflow for CPU but I get the same error as well.
I had a similar issue. In my case, I was trying to run luminoth inside a docker container. I resolved my issue by using nvidia-docker run instead of docker run.
Also, please check if you have compatible versions for cuda, cudnn, tensorflow-gpu - https://www.tensorflow.org/install/source#tested_build_configurations
If that doesn't work for you still, try seeing the list of GPU's you have available using nvidia-smi -q -g 0 -d UTILIZATION -l and set the device you want to use as visible using os.environ["CUDA_VISIBLE_DEVICES"]="0" https://stackoverflow.com/questions/37893755/tensorflow-set-cuda-visible-devices-within-jupyter
