search

GFNOrg / EB_GFN

Code for our paper "Generative Flow Networks for Discrete Probabilistic Modeling"
MIT License
73 stars 19 forks source link

Question about MH rejection rule #3

Closed violet-sto closed 1 year ago

violet-sto commented 1 year ago

Hi,

The code for compute MH is: update_dist = torch.distributions.Bernoulli(logits=lp_update + delta_logp_traj)

Is this inconsistent with the formula p=min(1, exp(lp_update+delta_logp_traj))