I'm using ML.NET for a value prediction application. Everything works, except I can't get it to use the LightGbm trainer algorithm. If I select all available trainers, it never tries LGBM. If I only select LGBM, I get the following error message:
at Microsoft.ML.AutoML.ConcatenateEntity.ToString() at Microsoft.ML.AutoML.SweepablePipeline.AppendEntity(Boolean allowSkip, Entity entity) at Microsoft.ML.ModelBuilder.AutoMLService.Experiments.RegressionExperiment.d__14.MoveNext() in //src/Microsoft.ML.ModelBuilder.AutoMLService/Experiments/RegressionExperiment.cs:line 109 End of stack trace from previous location where exception was thrown --- at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw() at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task) at Microsoft.ML.ModelBuilder.AutoMLEngine.d__21.MoveNext() in //src/Microsoft.ML.ModelBuilder.AutoMLService/AutoMLEngineService/AutoMLEngine.cs:line 169
I've used other NN products, and LGBM produces the best results for this application. Any ideas on how to get ML.NET to use it?
I've tried setting the training time to 6 hours (instead of the recommended 10 minutes) in hopes that it would pick LGBM as one of the trainers. It cycles through all of the other choices, but never uses LGBM. I've tried to train the net programmatically instead of through the UI, but there doesn't appear to be any option to pick a specific training algorithm.