bugfix for per-sample UQ

[ckhansen]

This is a small bugfix for UQ. When I was running AMS it was appearing to give me per-sample-dimension uncertainty values or maybe just multiple dimension uncertainty values. This just takes the max value if multiple uncertainties are output.

[github-actions]

Cpp-linter Review

Used clang-format v18.1.8

Click here for the full clang-format patch

diff --git a/src/AMSlib/ml/surrogate.cpp b/src/AMSlib/ml/surrogate.cpp
index 21babee..a1801e7 100644
--- a/src/AMSlib/ml/surrogate.cpp
+++ b/src/AMSlib/ml/surrogate.cpp
@@ -182 +182 @@ std::tuple<torch::Tensor, torch::Tensor> SurrogateModel::_evaluate(
-  
+
@@ -188 +188,2 @@ std::tuple<torch::Tensor, torch::Tensor> SurrogateModel::_evaluate(
-            "UQ model returned per-output uncertainty {}, reducing to per-sample by taking max",
+            "UQ model returned per-output uncertainty {}, reducing to "
+            "per-sample by taking max",

Have any feedback or feature suggestions? Share it here.

[lpottier]

LGTM. Thanks @ckhansen . Make sure to apply the clang-format recommendations before merging. Also make sure to rebase against develop (I just pushed a PR).

[lpottier]

@ckhansen I have push a commit with the correct clang-format. If tests are green, I will merge.

[ckhansen]

@lpottier Thanks for committing those changes!