4 ms·
Why won't you just try training on a smaller network? That way you 'force' the network to train on the small amount of filters you have right?
by Aardappelkroket 7y ago
Why won't you just try training on a smaller network? That way you 'force' the network to train on the small amount of filters you have right?
- sanxiyn 7y agoEmpirically, training smaller network fails. This paper is an attempt to explain why.