first install miniforge https://github.com/conda-forge/miniforge if not installed (or use existing conda with libmamba solver) then run `mamba env create -f env.yml` to create the environment (or env_aws.yml if not using GPU) then run `conda activate battery_minimal_gpu` or `(battery_minimal_cpu)` to activate the environment test command: `python bot/tariff_evaluate.py --class_name=StatefulRNNPolicy --data bot/data/validation_data.csv --output_file bot/results/statefulrnn_validation_results.csv` note: docker functionalities probably do not work training script is `exp_scripts/run_UNIFIED_experiment.py` and model parametered can be modified in the script the model is defined at `bot/policies/modules/stateful_rnn_model.py` the in-model battery environment is defined at `bot/policies/modules/battery.py`