configs/cluster/mlflow_logging.yaml

# @package _global_
# Configuration of mlflow logger. This is basic example of the usage.

trainer:
  # We rely on hydras dir management
  default_root_dir: .
  logger:
    _target_: ocl.utils.logging.ExtendedMLFlowLogger
    # Override if you want to have a different structure
    experiment_name: ${slice:${hydra:runtime.choices.experiment},"/",0}
    run_name: ${slice:${hydra:runtime.choices.experiment},"/","1:"}_${now:%Y-%m-%d_%H-%M-%S}

experiment:
  callbacks:
    checkpoint:
      _target_: pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint
    log_hydra_config:
      _target_: ocl.utils.logging.LogHydraConfigCallback
      hydra_output_subdir: ${hydra:output_subdir}
      # Add this in order to track parameters from the hydra config as hyperparameters
      additional_paths:
    log_model_summary:
      _target_: ocl.utils.logging.LogModelSummaryCallback

hydra:
  job:
    chdir: true