speechbrain.utils.check_yaml module

Tests for checking consistency between yaml files and their corresponding training scripts.

Authors

Mirco Ravanelli 2022
Andreas Nautsch 2022

Summary

Functions:

`check_module_vars`	Checks if the variables self.moduled.var are properly declared in the hparam file.
`check_yaml_vs_script`	Checks consistency between the given yaml file (hparams_file) and the script file.
`detect_script_vars`	Detects from the input script file (script_file) which of given variables (var_lst) are demanded.
`extract_patterns`	Extracts a variables from start_pattern to end_pattern.
`get_yaml_var`	Extracts from the input yaml file (hparams_file) the list of variables that should be used in the script file.

Reference

speechbrain.utils.check_yaml.get_yaml_var(hparam_file)[source]

Extracts from the input yaml file (hparams_file) the list of variables that should be used in the script file.

Parameters: hparam_file (path) – Path of the yaml file containing the hyperparameters.
Returns: var_list – list of the variables declared in the yaml file (sub-variables are not included).
Return type: list

speechbrain.utils.check_yaml.detect_script_vars(script_file, var_lst)[source]

Detects from the input script file (script_file) which of given variables (var_lst) are demanded.

Parameters

script_file (path) – Path of the script file needing the hyperparameters.
var_lst (list) – list of the variables declared in the yaml file.

Returns

detected_var – list of the variables detected in the script file.

Return type

list

speechbrain.utils.check_yaml.check_yaml_vs_script(hparam_file, script_file)[source]

Checks consistency between the given yaml file (hparams_file) and the script file. The function detects if there are variables declared in the yaml file, but not used in the script file.

Parameters

hparam_file (path) – Path of the yaml file containing the hyperparameters.
script_file (path) – Path of the script file (.py) containing the training recipe.

Returns

This function returns False if some mismatch is detected and True otherwise. An error is raised to inform about which variable has been declared but not used.

Return type

Bool

speechbrain.utils.check_yaml.extract_patterns(lines, start_pattern, end_pattern)[source]

Extracts a variables from start_pattern to end_pattern.

Parameters

lines (list) – List of strings to parse.
start_pattern (string) – String that indicated the start of the pattern.
end_pattern (string) – String that indicated the end of the pattern.

Returns

var_lst – List of variables detected.

Return type

list

speechbrain.utils.check_yaml.check_module_vars(hparam_file, script_file, module_key='modules:', module_var='self.modules.')[source]

Checks if the variables self.moduled.var are properly declared in the hparam file.

Parameters

hparam_file (path) – Path of the yaml file containing the hyperparameters.
script_file (path) – Path of the script file (.py) containing the training recipe.
module_key (string) – String that denoted the start of the module in the hparam file.
module_var (string) – String that denoted the start of the module in the script file.

Returns

This function returns False if some mismatch is detected and True otherwise. An error is raised to inform about which variable has been used but not declared.

Return type

Bool