Jerry Ling

another application has come up: https://arxiv.org/pdf/2001.05310 Basically, I'd like to add a term to the classifier loss function to encourate the model output to be de-correlated with some other variables

> if I'm understanding correctly, the problematic is not just about defining a new loss function as f(pred, y) (a function of only a model prediction and a target variable),...

>I'm unsure about the proper formulation for second order with Enzyme as I just revisited their docs. It behaves as expected with vanilla mse. But I wanted to illustrate how...

> if UnROOT can't handle their use case yet, use Python uproot via PyCall or PythonCall directly or use the ROOT wrapper? I'm fine with this. >if there aren't any...

Here's another manifestation of the same problem maybe, given: ```julia yscale=Makie.pseudolog10, yticks=10 .^ (0:7), limits=(-3, 3, 1, 10^7) ```  Note that `LogTicks` doesn't work (yet) because https://github.com/MakieOrg/Makie.jl/issues/4856

https://github.com/scikit-hep/awkward/blob/939b46c26ef73fb5900eb9d94537e68ff666e97b/src/awkward/_connect/cuda/cuda_kernels/awkward_reduce_min.cu#L72 is this the vanilla `atomicMin` from CUDA? It looks like this only handles integer? https://stackoverflow.com/questions/17399119/how-do-i-use-atomicmax-on-floating-point-values-in-cuda ## Edit: that's indeed the issue ```diff @@ -69,7 +75,7 @@ awkward_reduce_min_b( int64_t parent...

the problem is this code path can potentially launch more than 1 threads, we need a general solution anyway because we use this in more than one place. But I...

if you train a `machine`, and then: ```julia mach = restore!(serializable(mach_bdt)) ``` any subsequent call to `predict(mach, X)` would run on CPU not GPU. In this particular case the wrapper...

GPU training for XGBoost model is okay: https://github.com/JuliaAI/MLJXGBoostInterface.jl/issues/54 I think the problem is more that there's no way to move inference back to GPU -- I think in thsi case...

I think in this case the docs is at fault. It should say that identity only holds for `T` being a concrete type. edit: I meant to distinguish where a...