Text this: Benchmarking with a Language Model Initial Selection for Text Classification Tasks