deepreflect icon indicating copy to clipboard operation
deepreflect copied to clipboard
verified publisher icon evandowning

Nan value of lose and MSE

Open Zaibali9999 opened this issue 3 years ago • 2 comments

lose = nan - mse = nan while training the model and not improving the values

Zaibali9999 avatar Mar 16 '22 18:03 Zaibali9999

I also encountered this issue. This is due to the presence of a feature with a maximum value of 0, which leads to improper normalization that sets 0 as the denominator.

jimingming avatar May 11 '24 00:05 jimingming

Remove those feature which are missing while normalisation

On Sat, 11 May 2024, 03:53 jimingming, @.***> wrote:

I also encountered this issue. This is due to the presence of a feature with a maximum value of 0, which leads to improper normalization that sets 0 as the denominator.

— Reply to this email directly, view it on GitHub https://github.com/evandowning/deepreflect/issues/11#issuecomment-2105416042, or unsubscribe https://github.com/notifications/unsubscribe-auth/APUOQXHCISDZ64ZZP6YZ2KLZBVTYNAVCNFSM5Q4UVXO2U5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TEMJQGU2DCNRQGQZA . You are receiving this because you authored the thread.Message ID: @.***>

Zaibali9999 avatar May 12 '24 06:05 Zaibali9999