Hi, is there a way to tensorize T.init() into my own functions after I called decompose_reduction? There doesn’t seem to be an example about this. I think this would be really useful for me to initialize the result of the computation with the bias.
