List of available attributes in config file

The list of attributes correpsonds to the configurable variables in the ‘config.yaml’. The path of the ‘config.yaml’ file should be given as the one of the input parameters for genreat_modelcard(). The table below outlines the attributes that can be filled in the model card for the current version of dreams_mc. See the template of the config file here.

Attribute Name	Description
model_version	The version number of the model card.
logo_path	The path to the logo of the project or organisation.
dataset_name	Name of the dataset used for training the model.
num_target_class	Number of classes in the dataset under supervised setting.
ground_truth	Target label for training and validating the model under supervised setting.
split_ratio	The ratio for splitting dataset into training and validation and test set.
preprocess_steps	Names of the preprocess steps applied to the data.
model_type	Name of the model/architecture applied.
model_input	Input to the model (as a list if many input types) .
model_output	Model´s Output (as a list if many).
learning_rate	Learning rate used in training.
batch_size	Batch size of data .
additional_info	Additional information about the model (if required).
describe_overview	Overview of the model report.
describe_dataset	Description of the dataset used.
model_details	Description of the applied model.
limitation_details	Describing limitations about the model.
performance_comments	Describing about the model´s performance.
uncertainty_describe	Describing the uncertainty of the model´s performance.
data_figpath	Path to the data distribution figure (obtained from exploratory data analysis)
loss_figpath	Path to the training and validation loss figure(obtained after model training and validation)
acc_figpath	Path to the training and validatioin accuracy figure (obtained after model training and validation)
cm_figpath	Path to the confusion matrix (obtained after model training and validation)
uncertainty_figpath	Path to the figure depicting uncertainty estimation (obtained after model training and validation )
result_table_figpath	Path to the figure for result table (if necessary to display).