Neurips paper formula

[aced125]

Hi. Could you please explain how the log_prob function for the tensor ring induced prior works - mainly this line: https://github.com/insilicomedicine/TRIP/blob/master/core/learnable_priors/trip.py#L164

It seems to conflict with Algorithm 2 in the Neurips paper (https://arxiv.org/pdf/1910.13148.pdf), as in that algorithm, the log_probs are only calculated at the end, whereas in the code, it is updated at every iteration in the for loop.