If training a model with more than one spatial dimension, the visualization of the linear filter (space against time) does not produce recognizable results.
These changes allow to select the spatial dimension to plot against, while preserving previous behavior as this is implemented with optional parameters.
nirum
commented
5 years ago
If training a model with more than one spatial dimension, the visualization of the linear filter (space against time) does not produce recognizable results.
These changes allow to select the spatial dimension to plot against, while preserving previous behavior as this is implemented with optional parameters.