Normalized Configurations

Model configuration classes in 🤗 Transformers are not standardized. Although they provide an attribute_map attribute that enables to aleviate that, it does not make it easy to reason on common configuration attributes in the code. NormalizedConfig classes try to fix that by allowing access to the configuration attribute they wrap in a standardized way.

Base class

While it is possible to create NormalizedConfig subclasses for common use-cases, it is also possible to overwrite the original attribute name -> normalized attribute name mapping directly using the with_args() class method.

class optimum.utils.NormalizedConfig

< >

( config: PretrainedConfig allow_new: bool = False **kwargs )

Parameters

  • config (PretrainedConfig) — The config to normalize.

Handles the normalization of PretrainedConfig attribute names, allowing to access attributes in a general way.

Existing normalized configurations

class optimum.utils.NormalizedTextConfig

< >

( config: PretrainedConfig allow_new: bool = False **kwargs )

class optimum.utils.NormalizedSeq2SeqConfig

< >

( config: PretrainedConfig allow_new: bool = False **kwargs )

class optimum.utils.NormalizedVisionConfig

< >

( config: PretrainedConfig allow_new: bool = False **kwargs )

class optimum.utils.NormalizedTextAndVisionConfig

< >

( config: PretrainedConfig allow_new: bool = False **kwargs )