Skip to content

Set weight to 0.5 for non-supported layers in longitudinal weight correction in AF3 simulation

Layers not supported by the neural network will get a weight=0.5.

This is to support some corner cases where we ask the TFCSPredictExtrapWeights class for an extrapolation weight for a non-supported layer.

@mfauccig @cyoung

Edited by Jonathan Bossio

Merge request reports