MLContext seed has not impact on evaluating regression metrics

Hi, I have problems getting deterministic results using ml.net regression evaluation.
Each time I run the following code sample from this guthub repo, which is using a seed, the regression metrics calculated are slightly different. Shouldnt they be always the same, to provide compareable results?

Example i used to reproduce the problem:
https://github.com/dotnet/samples/tree/main/machine-learning/tutorials/TaxiFarePrediction

I also updated to the latest ml.net (1.7) version.
Looking forward for your reply!

Best regards flotschi!