pip install -r requirements.txt

Directory for saving and unloading datasets while the benchmark is running
```bash
export DATASETSROOT=<enter absolute path>

python workloads/load_datasets.py

For runs all workloads:

python benchmarks/svm_workload_run.py

For runs of the selected workload:

python benchmarks/svm_workload_run.py --workload a9a

You can choose library: sklearn, cuml, thundersvm, sklearn-intelex For runs of the selected library. By default using sklearn with oneDAL optimizations (sklearn-intelex). Example for thundersvm library:

python benchmarks/svm_workload_run.py --library thundersvm

NOTE: for thundersvm/cuml runs need thundersvm/cuml library. Can you download with help pip or conda

You can also choose a task for the svm: svc, svc_proba, svr. By default svc