Getting loss value for testing set

[VittoriaOssanna]

Hello everyone,

I am having troubles getting the loss from a test set. My dataset is divided into train and test through sklearn. I understand that I can use model.predict(test_set), but I was wondering if there is any direct way of getting the loss of the test batch along with every prediction. I implemented it myself but I am sure enough that is not the smartest way.

Can somebody help me?
Thank you!

[MilesCranmer]

Maybe this?

from sklearn.metrics import mean_squared_error as mse

y_pred_test = model.predict(x_test)
mse(y_pred_test, y_test)

[VittoriaOssanna]

What if I used LogitDistLoss, L2DIstLoss and L1DitstLoss? Is there any quick way to compute the loss?

[MilesCranmer]

Hm, maybe try

summed_loss = jl.seval(
"""
(pred, target) -> begin
    loss = L2DistLoss()
    sum(loss(pred[i], target[i]) for i in eachindex(pred, target))
end
"""
)
summed_loss(y_pred_test, y_test)