ignore-cols for mlnet not working as it includes all other input columns to features

[cyberkoolman]

System Information (please complete the following information):

• OS & Version: Windows 10 & Windows 11
• ML.NET Version: ML.NET v2.0.1
• .NET Version: .NET 7.0

Describe the bug
Using the mlnet cli on wikiDetoxAnnotated40kRows.tsv which has the columns of Label, rev_id, comment, year, logged_in, ns, sample, split, wanted to pick only comment as the input feature. Currently, through CLI, have to use --ignore-cols parameter and resulted the training pipeline to include all the columns to feature. I had to resort to VS ModelBuilder for go around this issue for now.

To Reproduce
Execute cli as below
mlnet classification --dataset .\wikiDetoxAnnotated40kRows.tsv --label-col Label --has-header true --ignore-cols rev_id year logged_in ns sample split --name Sentiment

Expected behavior
I would expect the features only contain "comment" field

Screenshots, Code, Sample Projects
Resulting training pipeline as below

Additional context

1. Shouldn't this be easier if providing the option to choose only relevant column, instead of long list of ignore-columns.
2. Documentation says ignore-columns, not matching to ignore-cols. [https://learn.microsoft.com/en-us/dotnet/machine-learning/reference/ml-net-cli-reference#ignore-columns]

[KeithT]

I have this issue with the model builder 16.13.9.2235601, it ignores the ignore enum and trains on all the columns in the dataset.

[LittleLittleCloud]

@KeithT @cyberkoolman Thanks for reporting that, we are looking at it now.

@KeithT Can you update your model builder to latest version and see if the issue is still there?

[KeithT]

@LittleLittleCloud I updated to 17.14.4.2312404 and the ignore columns does now work. (Thanks). It still falls over when it's finished training and tries to write code though, as I reported here:
#2541
I've posted a screenshot again below.

[LittleLittleCloud]

@cyberkoolman @KeithT thanks for reporting, we'll push a fix pretty soon

[LittleLittleCloud]

Should be fixed in 16.17.17. Please reopen it if this error appear again