You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think it was a joke to refer to those learners that had this relatively heavyweight MKL library as a dependency for one reason or another.
I like Microsoft.ML.MklTrainers or Microsoft.ML.MklComponents or Microsoft.ML.Mkl or something like this. (If possible Trainers would be good, but there's more in there that depends on MKL than merely trainer estimators -- so maybe the more general name would be good).
"Hardware accelerated" is far too prone to misunderstanding in my opinion. Without knowing what I know, I think if I were to look at this library, and I saw, "hardware accelerated," my natural thought would be "OK, so, they are saying the rest the stuff isn't hardware accelerated." And, that would be incorrect while also making us look kind of silly.
I see. Well, I'm not sure that's a blocker. We have after all the "transforms" assembly without the suggestion that it refers to all transforms. (Indeed it does not.) How about Microsoft.ML.Mkl.Components then, anyway?
The name
HalLearners
isn't a great public name. We should come up with a better name.Current issues:
Hal
mean?Possible suggestions to get the naming started:
Microsoft.ML.HardwareAccelerated
Microsoft.ML.HardwareAccelerated.Components
Microsoft.ML.Mkl
Microsoft.ML.Mkl.Redist
. this would be an analog toML.TensorFlow
andML.TensorFlow.Redist
)Microsoft.ML.Mkl.Components
@TomFinley @sfilipi @Ivanidzo4ka @shauheen @yaeldekel
The text was updated successfully, but these errors were encountered: