Commit ff8cee2

committed

Ver 0.3

1 parent a516d79 commit ff8cee2Copy full SHA for ff8cee2

File tree

-2

lines changed

-2

lines changed

Lines changed: 5 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -38,4 +38,8 @@ import torchbnn`
`38`	`38`	`* convert_model(nonbayes_to_bayes, bayes_to_nonbayes) is also changed with prior_sigma instead of prior_log_sigma.`
`39`	`39`	`* Modules(BayesLinear, BayesConv2d, BayesBatchNorm2d) : Base initialization method is changed to the method of Adv-BNN from the original torch method.`
`40`	`40`	`* functional : bayesian_kl_loss is changed similar to ones in torch.functional`
`41`		`-* loss : BKLLoss is added based on bayesian_kl_loss similar to ones in torch.loss`
	`41`	`+* loss : BKLLoss is added based on bayesian_kl_loss similar to ones in torch.loss`
	`42`	`+`
	`43`	`+### Version 0.3`
	`44`	`+* bayesian_kl_loss/BKLLoss returns tensor.Tensor([0]) as default`
	`45`	`+ * In the previous version, bayesian_kl_loss/BKLLoss returns 0 of int type if there is no Bayesian layers. However, considering all torch loss returns tensor and .item() is used to make them to int type, they are changed to return tensor.Tensor([0]) if there is no Bayesian layers.`

Lines changed: 2 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -33,7 +33,8 @@ def bayesian_kl_loss(model, reduction='mean', last_layer_only=False) :`
`33`	`33`
`34`	`34`	`"""`
`35`	`35`
`36`		`- kl_sum = 0`
	`36`	`+ kl = torch.Tensor([0])`
	`37`	`+ kl_sum = torch.Tensor([0])`
`37`	`38`	`n = 0`
`38`	`39`
`39`	`40`	`for m in model.modules() :`

Comments

(0)