if we have a loss like this plot, is it kind of underfitting or goodfitting? enter image description here

Error results: enter image description here

Training Process: enter image description here

  • 3
    $\begingroup$ The loss of the first training batch probably is an outlier that does not let you see the details of the curves past it. I suggest that you remove it and check how it looks. $\endgroup$
    – noe
    Commented May 2, 2023 at 7:37
  • $\begingroup$ Your y axis scale is incorrect because what you want to measure are values very close to 0. You should plot log(loss) instead of loss. The same is true for validation. Log makes it way more readable for values close to 0. $\endgroup$
    – Ubikuity
    Commented May 3, 2023 at 11:50


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.