Why not directly infer the value of x-based distribution?

[CXX1113]

Hello, thank you very much for your excellent work.
In deepscm-master\deepscm\experiments\morphomnist\nf\conditional_image_flow_sem.py line 132-140
Why not directly infer the value of x-based distribution like the function of infer_intensity_base or infer_e_t?
In your code, you still use "x_base_dist = Normal(self.x_base_loc, self.x_base_scale)" to generate the inference value of x_based.