The question of training.py

[5466a]

What is the function of [ Computing log-likelihood on test set ] ?

[riannevdberg]

Hi there,

It computes an upper bound on the negative log likelihood (in bits per dimension) using importance sampling. I hope that helps.

[5466a]

Uh... I understand that, but what I want to ask is what is the function or significance of computing this?

[bgroenks96]

It can be used a metric for comparison between models on the same test set.

[5466a]

Ok, thank you very much | | 宫文涛 | | ***@***.*** | 签名由网易邮箱大师定制 On 9/2/2021 15:23，Brian ***@***.***> wrote： It can be used a metric for comparison between models on the same test set. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[5466a]

It can be used a metric for comparison between models on the same test set.

Hello, I have a new question: how do I set the size of S and MB on other datasets?