How to use the tcav.cav.get_or_train_cav function in tcav

To help you get started, we’ve selected a few tcav examples, based on popular ways it is used in public projects.

Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.

github amiratag / ACE / ace.py View on Github external
Args:
      c: conept name
      r: random concept name
      bn: the layer name
      act_c: activation matrix of the concept in the 'bn' layer
      ow: overwrite if CAV already exists
      directory: to save the generated CAV

    Returns:
      The accuracy of the CAV
    """
    if directory is None:
      directory = self.cav_dir
    act_r = self._random_concept_activations(bn, r)
    cav_instance = cav.get_or_train_cav([c, r],
                                        bn, {
                                            c: {
                                                bn: act_c
                                            },
                                            r: {
                                                bn: act_r
                                            }
                                        },
                                        cav_dir=directory,
                                        overwrite=ow)
    return cav_instance.accuracies['overall']

tcav

Testing with Concept Activation Vectors code

Apache-2.0
Latest version published 3 years ago

Package Health Score

51 / 100
Full package analysis

Similar packages