config

.

Check-out the hydra docs & omegaconf docs for more information on how structured configurations work and how to best utilize them.

class common.config.BaseSubtaskConfig(output_dir='${hydra:runtime.output_dir}', data_dir='${oc.env:AI_REPO_PATH}/data/', device='cpu', seed=0)

Bases: object

Base subtask config.

Parameters:

• output_dir (str, default: '${hydra:runtime.output_dir}') – Path to the subtask output directory. Every artifact generated during the subtask will be stored in this directory.

• data_dir (str, default: '${oc.env:AI_REPO_PATH}/data/') – Path to the data directory. This directory is shared between task runs. It can be used to store datasets, pre-trained models, etc.

• device (str, default: 'cpu') – Computing device to use for large matrix operations.

class common.config.BaseHydraConfig(defaults=<factory>, mode=RunMode.MULTIRUN, searchpath=<factory>, run=<factory>, sweep=<factory>, hydra_logging='???', job_logging='???', sweeper='???', launcher='???', callbacks=<factory>, help=<factory>, hydra_help=<factory>, output_subdir='.hydra', overrides=<factory>, job=<factory>, runtime=<factory>, verbose=False)

Bases: Config