braindecode.visualization.run_with_activation_substitution#

braindecode.visualization.run_with_activation_substitution(model, x, layer, substitute_fn)[source]#

Run model(x) after replacing layer’s output.

Parameters:

  • model (Module) – Model to run.

  • x (torch.Tensor) – Model input.

  • layer (Module) – Submodule whose output is replaced.

  • substitute_fn (callable) – Called with the layer output and must return its replacement.

Returns:

The model output after substitution.

Return type:

Any

Raises:

TypeError – If substitute_fn returns None, which PyTorch interprets as keeping the original output.