FB13516799: Training Tabular Regression ML Models on large datasets in Xcode 15 continuously "Processing"

Hi,

In Xcode 14 I was able to train linear regression models with Create ML using large CSV files (I tested on about 30000 items and 5 features):

However, in Xcode 15 (I tested on 15.0.1 and 15.1), the training continuously stays in the "Processing" state:

When using a dataset with 900 items, everything works fine.

I filed a feedback for this issue: FB13516799.

Does anybody else have this issue / can reproduce it?

I've tried to look into this further with the Console app, and it seems that everything might be working but very slowly due to the large dataset. However, I don't remember this process being that slow on Xcode 14.

The only reports I found were some cpu usage and wakeups diagnostic reports.

After leaving it process the data for a while, I eventually got some <private> class has successfully loaded the model at <private>. and The optimizer converged after ... iterations. streamed by the MLRecipeExecutionService (CoreML) and IntelligencePlatformComputeService (CoreML) processes after 5 minutes. But after 30 minutes the training is still not finished.

Another log message I found is the following:

<private> class was unable to load the model at <private> with error: <private>; The model loader is going to use another class.

I meant tabular regression models.

I'm having the same issue with just 4573 items (5kb).

FB13516799: Training Tabular Regression ML Models on large datasets in Xcode 15 continuously "Processing"
 
 
Q