src.loggers package

Submodules

src.loggers.FileLogger module

class src.loggers.FileLogger.FileLogger(log_dir: str, experiment_name: str)

Bases: BaseLogger

Create file logger using Python’s default logging mechanism.

Parameters

• log_dir (str) – Log directory

• experiment_name (str) – Name of the experiment in question.

log(data)

Log data

Parameters

data (dict) – Data to log.

src.loggers.TensorboardLogger module

Wrapper around TensorBoard for logging. Unused for WandB, but should work.

class src.loggers.TensorboardLogger.TensorboardLogger(log_dir: str, experiment_name: str)

Bases: BaseLogger

TBLogger Instance

TensorBoard Logger

Parameters

• log_dir (str) – Log directory

• experiment_name (str) – Experiment Name

log(metric_data, ep_num)

Log metric. Silently fails if something goes wrong.

Parameters

• metric_name (dict) – Dictionary of metric_name to metric_value

• ep_num (int) – Metric location ( x-value )

src.loggers.WanDBLogger module

Weights and Biases Logging.

class src.loggers.WanDBLogger.WanDBLogger(api_key: str, project_name: str)

Bases: BaseLogger

Wandb Logger Wrapper.

Create Weights and Biases Logger

Parameters

• api_key (str) – api key (DO NOT STORE IN REPO)

• project_name (str) – project name

log(data)

Log metrics to WandB, using names present in dict.

Parameters

data (dict) – Dict to log

src.loggers.base module

Logger Base Class.

class src.loggers.base.BaseLogger(log_dir, exp_name)

Bases: ABC

Base Logger.

Initialize logger

Parameters

• log_dir (str) – Log directory

• exp_name (str) – Experiment Name

log(logging_data)

Log data

Parameters

logging_data (dict) – Data to log

Module contents

Logger definitions to send data to some place for later.