Pretty hacky way of doing this, but at least consistent with how I did the nonlinear scviv2. Let me know what you think. (perhaps redun is better in this aspect too).
Note: the scVI baselines have redundant training which can be really bad. Would be good to figure out a way to factor this computation out @martinkim0
Pretty hacky way of doing this, but at least consistent with how I did the nonlinear scviv2. Let me know what you think. (perhaps redun is better in this aspect too).
Note: the scVI baselines have redundant training which can be really bad. Would be good to figure out a way to factor this computation out @martinkim0