hyperparam_reporter

class aitoolbox.experiment.result_reporting.hyperparam_reporter.HyperParamSourceReporter(project_name, experiment_name, experiment_timestamp, local_model_result_folder_path)[source]

Bases: object

Writer of selected hyperparameters to human-readable text file on disk

Parameters
  • project_name (str) – root name of the project

  • experiment_name (str) – name of the particular experiment

  • experiment_timestamp (str) – time stamp of the training start

  • local_model_result_folder_path (str) – root local path where project folder will be created

save_hyperparams_to_text_file(hyperparams, sort_names=False)[source]

Save hyperparameters dict into text file on disk

Parameters
  • hyperparams (dict) – hyper-parameters listed in the dict

  • sort_names (bool) – should presented hyper-param names be listed alphabetically

Returns

path to the saved hyper-param text file

Return type

str

copy_to_cloud_storage(local_hyperparams_file_path, cloud_saver, file_name=None)[source]

Copy saved text local file into cloud storage

Parameters
Returns

path where the file was saved in the cloud storage

Return type

str

save_experiment_python_file(hyperparams)[source]

Saves the python experiment file to the project folder

Python experiment file is file in which the main training procedure is defined. File from which the TrainLoop is executed

Parameters

hyperparams (dict) – hyper-parameters listed in the dict. In order for this function to work, the dict needs to include experiment_file_path key.

Returns

path to the saved main python experiment file

Return type

str

save_experiment_source_files(hyperparams)[source]

Saves all the experiment source files into single source code zip

Parameters

hyperparams (dict) – hyper-parameters listed in the dict. In order for this function to work, the dict needs to include source_dirs_paths key.

Returns

path to the saved experiment source code zip

Return type

str