This package provides convenient metrics for testing generated time-series data. It consists of two parts: standalone metrics that measure the distance between distributions as well as metrics that measure performance in downstream classification tasks. Both parts can be used independently.
The installation consists of two parts:
- Installing dependencies using
conda/mamba - Installing package using pip.
Supported way of installing dependencies is through conda/mamba.
Please substitute new_env_name with your prefered conda environment name
conda env create -n new_env_name -f environment.yml
conda activate new_env_namepip install git+https://gitlab1.ptb.de/qai-labs/generative-models/benchmarkpython -m benchmark.cli --helpThe documentation is available at https://example.com/docs
Please create issues in this repositories in case you have questions concerning the usage of the package.
State if you are open to contributions and what your requirements are for accepting them.
To start tinkering just clone repository and do edit installation with pip:
git clone https://gitlab1.ptb.de/qai-labs/generative-models/benchmark
cd benchmark
python -m pip install -e .