Hengyuan cloud (Gpushare)_JupyterLab/TensorBoard usage problem? Skill Big Show 4!

Article Source | Hengyuan Cloud Community

Original address | [Tips - Tools]

---

1. How to use JupyterLab/TensorBoard?

Refer to JupyterLab documentation and TensorBoard documentation.

2. What should I do if JupyterLab cannot be opened or the response is slow?

If the response of JupyterLab page is very slow or cannot be opened, you can restart JupyterLab with the following command after logging in to the instance

 supervisorctl restart jupyterlab

It is not recommended to directly run programs that consume more resources through the JupyterLab page, because this program is likely to cause JupyterLab to be unresponsive or run slowly. It can be placed in the background of the instance through the tmux tool.

3. What should I do if the tensorboard cannot be opened or the response is slow?

If the response of the tensorboard page is extremely slow or cannot be opened, you can log in to the instance and run the following command to restart tensorboard

 supervisorctl restart tensorboard

4. How to install jupyterlab with custom image?

After customizing the image login instance, execute the following command to complete the installation of jupyterlab. After the installation is complete, you can open the jupyterlab page through the console

 curl -OL "https://download.gpushare.com/download/custom_install"
chmod u+x ./custom_install
./custom_install jupyterlab

5. How to install tensorboard in a custom image?

After customizing the image to log in to the instance, execute the following command to complete the installation of tensorboard. After the installation is complete, you can open the tensorboard page through the console

 curl -OL "https://download.gpushare.com/download/custom_install"
chmod u+x ./custom_install
./custom_install tensorboard

6. What if JupyterLab needs to enter a password?

Execute the jupyter server list command through the terminal to get the login Token of JupyterLab.

The token obtained from the execution result is 3fq593blw4afqjtqgdp3ldk5 as follows.

 root@I15b96311d0280127d:~# jupyter server list
Currently running servers:
http://0.0.0.0:8888/?token=3fq593blw4afqjtqgdp3ldk5 :: /

7. TensorFlow training reports ptxas fatal error

When training with TensorFlow 2.4 For CUDA 11.0 on RTX 3000 series graphics cards, the following warning appears.

 W tensorflow/stream_executor/gpu/asm_compiler.cc:235] Your CUDA software stack is old. We fallback to the NVIDIA driver for some compilation. Update your CUDA version to get the best performance. The ptxas error was: ptxas fatal : Value 'sm_86' is not defined for option 'gpu-name'

The reason is that this version of the PTX compiler does not support the 8.6 compute capability. This error is a warning and does not affect normal training.

This warning can be turned off with os.environ['TF_CPP_MIN_LOG_LEVEL'] = "2".

Although this problem does not affect training, the performance will be reduced. It is recommended to create an image of TensorFlow 2.5 For CUDA 11.2, which will not have this problem.