Installing Python Libraries in the daphne-dev
Container
The daphne-dev
container (see GettingStarted) already contains all required dependencies for running DAPHNE. However, there can be reasons to install additional Python libraries inside the container, e.g.:
- To use/test DaphneLib's data exchange with TensorFlow and PyTorch.
DaphneLib, DAPHNE's Python API, supports the efficient data exchange with widely-used Python libraries like numpy, pandas, TensorFlow, and PyTorch.
Numpy and pandas are required for DaphneLib.
Thus, they are already installed in the
daphne-dev
container. In contrast to that, TensorFlow and PyTorch are optional for DaphneLib; if these libraries are not installed on the system, DaphneLib cannot exchange data with them, but all remaining features still work. Likewise, the test cases related to the data exchange with TensorFlow and PyTorch will only run if these libraries are installed. As TensorFlow and PyTorch would increase thedaphne-dev
container size by several gigabytes, they are not included in the container. - To add support for additional Python libraries in DaphneLib. For instance, while implementing efficient data exchange with these additional libraries.
- To build integrated data analysis pipelines involving additional Python libraries. For instance, for experiments.
Installing Additional Python Libraries
Additional Python libraries are best installed in a Python virtual environment inside the daphne-dev
container.
To that end, execute the following commands inside the container:
Create a Python virtual environment and activate it:
sudo apt update
sudo apt install python3.12-venv
python3 -m venv daphne-venv
source daphne-venv/bin/activate
Here, we call the virtual environment daphne-venv
.
Feel free to choose a different name.
Install the desired Python libraries using pip
:
For instance, if you want to use/test DaphneLib's efficient data transfer with widely-used Python libraries like numpy, pandas, TensorFlow, and PyTorch, install the following libraries. Feel free to install any library you like.
The libraries you install that way will be stored in the daphne-venv
directory on the host and, thus, keep existing after you shut down the container.
Don't Forget
Every time you enter the daphne-dev
container, make sure to activate the Python virtual environment again: