CLI

rl8’s command-line interface (CLI). A selection of tools to support common workflows for reinforcement learning with rl8.

/home/runner/work/rl8/rl8/.tox/docs/lib/python3.10/site-packages/mlflow/pyfunc/model.py:175: UserWarning: [31mType hint used in the model's predict function is not supported for MLflow's schema validation. Type hints must be wrapped in list[...] because MLflow assumes the predict method to take multiple input instances. Specify your type hint as `list[dict[str, typing.Any]]` for a valid signature. Remove the type hint to disable this warning. To enable validation for the input data, specify input example or model signature when logging the model. [0m
  func_info = _get_func_info_if_type_hint_supported(predict_attr)
/home/runner/work/rl8/rl8/.tox/docs/lib/python3.10/site-packages/mlflow/pyfunc/model.py:175: UserWarning: [31mType hint used in the model's predict function is not supported for MLflow's schema validation. Type hints must be wrapped in list[...] because MLflow assumes the predict method to take multiple input instances. Specify your type hint as `list[dict[str, typing.Any]]` for a valid signature. Remove the type hint to disable this warning. To enable validation for the input data, specify input example or model signature when logging the model. [0m
  func_info = _get_func_info_if_type_hint_supported(predict_attr)
usage: rl8 [-h] {train} ...

positional arguments:
  {train}
    train     Training command to train a policy using the trainer interface.
              Provides a common training workflow that satisfies most use
              cases. Includes training from a config file, MLflow
              experiment/run setup, training validation, policy saving, and
              more.

options:
  -h, --help  show this help message and exit

Training

/home/runner/work/rl8/rl8/.tox/docs/lib/python3.10/site-packages/mlflow/pyfunc/model.py:175: UserWarning: [31mType hint used in the model's predict function is not supported for MLflow's schema validation. Type hints must be wrapped in list[...] because MLflow assumes the predict method to take multiple input instances. Specify your type hint as `list[dict[str, typing.Any]]` for a valid signature. Remove the type hint to disable this warning. To enable validation for the input data, specify input example or model signature when logging the model. [0m
  func_info = _get_func_info_if_type_hint_supported(predict_attr)
/home/runner/work/rl8/rl8/.tox/docs/lib/python3.10/site-packages/mlflow/pyfunc/model.py:175: UserWarning: [31mType hint used in the model's predict function is not supported for MLflow's schema validation. Type hints must be wrapped in list[...] because MLflow assumes the predict method to take multiple input instances. Specify your type hint as `list[dict[str, typing.Any]]` for a valid signature. Remove the type hint to disable this warning. To enable validation for the input data, specify input example or model signature when logging the model. [0m
  func_info = _get_func_info_if_type_hint_supported(predict_attr)
usage: rl8 train [-h] [-f FILE] [--experiment-name EXPERIMENT_NAME]
                 [--max-steps MAX_STEPS] [--save SAVE]
                 [--steps-per-eval STEPS_PER_EVAL]

options:
  -h, --help            show this help message and exit
  -f FILE, --file FILE  Train config file to build the trainer from.
  --experiment-name EXPERIMENT_NAME
                        MLflow experiment name to organize runs under.
                        Defaults to the environment's fully qualified name.
  --max-steps MAX_STEPS
                        Max trainer steps (each trainer step = number of
                        environments * horizon) before stopping.
  --save SAVE           Directory to save the trained policy to.
  --steps-per-eval STEPS_PER_EVAL
                        Number of trainer steps for each trainer eval.