1. Use TensorBoard
You can use TensorBoard in the project IDE environment.
TensorBoard provides visualizations and tools for machine learning experiments. Try it out with the guide below.
This guideline was written based on the GPU Jupyter container.
How to use
On the Yennefer Studio project details page, click "Activate Server > Run Server" to run the IDE.
After running the server, click Terminal in the Launcher.
Enter the following in the terminal in order.
pip install tensorflow pip install jupyter-server-proxy npm install -g n export N_PREFIX=/opt/conda/ n stable pip install git+https://github.com/twalcari/jupyterlab_tensorboard.git
4. When the installation is complete, close the window and deactivate the server. 5. Activate the server again and run the server.
6.Create a notebook to use TensorBoard
7.After writing the code as above, proceed with learning.
8. When training, logdir and tensorboard_callback variables must be added. (Reference code: https://www.tensorflow.org/tensorboard/tensorboard_in_notebooks)
9. When learning is completed normally, a logs folder is created.
10. Check hub0/user/X/Y/ in the address bar at the top.
11. Execute the code referring to hub0/User/X/Y in the address bar at the top.
Enter numbers in X and Y, and you can write as it is otherwise.
12. Enter %tensorboard --logdir logs
code to run TensorBoard..
Last updated