mapping_networks.models.ForwardResult¶

class mapping_networks.models.ForwardResult(predictions: Any, generated_parameters: ParameterTree, latent_vectors: dict[str, Tensor])¶

Structured output from MappingModel.forward().

Carries autograd-live tensors so downstream losses and the trainer can access generated parameters and latent vectors without positional argument gymnastics.

predictions¶

Raw output from the target model forward pass.

Type:: Any

generated_parameters¶

The ParameterTree produced by the generator.

Type:: mapping_networks.runtime.parameter_tree.ParameterTree

latent_vectors¶

Mapping from stable latent names to their trainable tensors.

Type:: dict[str, torch.Tensor]

__init__(predictions: Any, generated_parameters: ParameterTree, latent_vectors: dict[str, Tensor]) → None¶

Methods

__init__(predictions, generated_parameters, ...)

Attributes

`predictions`
`generated_parameters`
`latent_vectors`