Skip to content

Conversation

@BernhardAhrens
Copy link
Collaborator

@BernhardAhrens BernhardAhrens commented Dec 15, 2025

extra_loss function can be defined like so. Here steering rb to 5

extra_loss = function (ŷ)
    (; a=sum((5.0 .- ŷ.rb) .^ 2))
end

missing => might be a  pre-processing we do for the user
…rameters to 0.5 between min max

- hybrid_name fills up the disk space when experiementing
- log10 yscale breaks with NSE metric comes from negative to positive
@BernhardAhrens
Copy link
Collaborator Author

BernhardAhrens commented Dec 15, 2025

Working example in projects/book_chapter/example_synthetic.jl for both

… IDE has autoformat on save on with with other formatting rules

function compute_loss(ŷ, y::AbstractDimArray, y_nan::AbstractDimArray, targets, loss_spec, agg::Function)
losses = [_apply_loss(ŷ[k], y[col = At(k)], y_nan[col = At(k)], loss_spec) for k in targets]
losses = [_apply_loss(ŷ[k], y[col=At(k)], y_nan[col=At(k)], loss_spec) for k in targets]
Copy link
Member

@lazarusA lazarusA Dec 15, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

in #157 I did use col to match KeyArray syntax, and this will work for now, but note that in general DimArrays / YAXArrays will not have this name. We will need to discuss this further.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe we just rename whatever the user gives and print something original axis x renamed to row, other axis untouched

@lazarusA lazarusA changed the title [wip] GenericHybrid with DimensionalData and extra_loss function GenericHybrid with DimensionalData and extra_loss function Dec 16, 2025
@lazarusA lazarusA merged commit 14ab9c7 into main Dec 16, 2025
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants