Right now the visualizer accepts a variant tuple as an input, but that causes the visualizer to regenerate the encoding just to visualize. This is not ideal because it takes more time and the variant tuple is not always accessible. It's better if the visualizer is able to accept the encoded tensor as an input and directly visualize that.
Right now the visualizer accepts a
varianttuple as an input, but that causes the visualizer to regenerate the encoding just to visualize. This is not ideal because it takes more time and the variant tuple is not always accessible. It's better if the visualizer is able to accept the encoded tensor as an input and directly visualize that.