EnglishTextClassification error after successful training: Model has not been trained

Help AI Center
,
1

I am setting up a functioning project skeleton that I can improve later with real data. I have trained with the bare minimum for the pipeline to successfully train (5 cases per class). Why am I getting an error that the model has not been trained when the Pipeline tells me it has been? Do I need to include more training cases?

My data is:

input,target
this is a test,test
and this is a test,test
this is also a test,test
this is a bloody test for your damn 5 min,test
this is another test to meet the 5 min,test
this is not a test,notest
this certainly is not a test,notest
a test? No this can never be a test,notest
this is not a test at all ffs,notest
this isnt either. So few tests needed,notes

The error I get when trying to deploy an ML Skill for the ML Package I just trained:

File “/microservice/main.py”, line 12, in init
raise UiPathUsageException(_UNTRAINED_HELP)
aiflib.logger.UiPathUsageException:
This generic Text Classification AutoML model has not
been trained. Use AI Fabric to train this model on your own data. The model
will read all csv files in the directory recursively. Each file simpy must
have a header (the first line of the csv file) “input,target” and the
subsequent lines are: , pairs.

The Pipeline tells me the ML Package was successfully trained. The files I get in my training pipeline:
__results.json (380 Bytes) train-report.pdf (58.6 KB) b14a42b0-8b19-4fe2-b18d-fe1430d6be9b.txt (13.0 KB)
learning-rate-finder

If you need any more information let me know

2

Hi @David_Hampton
What is the exact version you are trying to deploy? I suspect you are not selecting the right minor version (the one generated by the pipeline), you need to deploy version X.1 (or more) not X.0.
Jeremy