nll computation

test_inputcomplexity.py

@@ -33,6 +33,7 @@ def KL_div(mu,logvar,reduction = 'none'):
 
 def store_NLL(x, recon, mu, logvar, z):
     with torch.no_grad():
+        sigma = torch.exp(0.5*logvar)
         b = x.size(0)
         target = Variable(x.data.view(-1) * 255).long()
         recon = recon.contiguous()
@@ -41,7 +42,7 @@ def store_NLL(x, recon, mu, logvar, z):
         log_p_x_z = -torch.sum(cross_entropy.view(b ,-1), 1)
 
         log_p_z = -torch.sum(z**2/2+np.log(2*np.pi)/2,1)
-        z_eps = z - mu
+        z_eps = (z - mu)/sigma
         z_eps = z_eps.view(opt.repeat,-1)
         log_q_z_x = -torch.sum(z_eps**2/2 + np.log(2*np.pi)/2 + logvar/2, 1)
 
@@ -245,4 +246,4 @@ def compute_NLL(weights):
             break
     difference_ood = np.asarray(difference_ood)
 
-    np.save('./array/complexity/difference_ood.npy', difference_ood)
+    np.save('./array/complexity/difference_ood.npy', difference_ood)