1. Log in to the cloud computing server

2. Download and install Anaconda because the r&d network cannot access the Internet, you can only download resources from the internal server

cd
mkdir local && cd localwget http://...... Anaconda/archive/Anaconda3-2019.03 - Linux - x86_64. Sh# your company's FTP server addressBash Anaconda3-2019.03 - Linux - x86_64. ShCopy the code

3. 

channels: - https://...... /anaconda/cloud/pytorch - https://...... /anaconda/cloud/menpo - https://...... /anaconda/cloud/bioconda - https://...... /anaconda/cloud/conda-forge - https://...... /anaconda/pkgs/main - https://...... /anaconda/pkgs/free - https://...... /anaconda/anthol show_channel_urls: yes ssl_verify:false
Copy the code

4.
This blog

5. 

Importerror: / lib64 / libm. So. 6: version'GLIBC_2. 23' not found (required by /home/...
Copy the code

Need to lower version:

Conda create --name tensorflow-gpu python=3.6source activate tensorflow-gpu
Copy the code

Do not discard the source in front of the command line (tensorflow-GPU). You can perform subsequent operations only in the tensorflow-GPU environment

conda search opencv

Opencv is specified as 3.4.2. As of 201907 OpencV has been updated to 4.1.0, but I feel the previous version is better

Conda install opencv = = 3.4.2

Install Tensorflow-GPU and Keras etc

conda install tensorflow-gpu keras joblib

conda install pytorch scikit-learn matplotlib

6. Finally, don’t forget to configure the environment to activate TensorFlow-GPU in the. Bashrc file

source activate tensorflow-gpu

7. Remind virtual desktop not to shut down!