This appears to be a typo. I think the correct behavior is that GoodTuring should take into account a vocab and encoding if given.
Side note: How do you run the tests in this repo?
I tried evaluating tests/test_ngram.py in a REPL but got an error about attempted relative import with no known parent package. I tried running python -m unittest but got a ModuleNotFoundError: No module named 'librosa.core.time_frequency'. I resorted to python -c 'from numpy_ml.tests import test_ngram; test_ngram.some_test(). Noting that there's not a test for GoodTuring yet.
This appears to be a typo. I think the correct behavior is that GoodTuring should take into account a vocab and encoding if given.
Side note: How do you run the tests in this repo?
I tried evaluating
tests/test_ngram.pyin a REPL but got an error aboutattempted relative import with no known parent package. I tried runningpython -m unittestbut got aModuleNotFoundError: No module named 'librosa.core.time_frequency'. I resorted topython -c 'from numpy_ml.tests import test_ngram; test_ngram.some_test(). Noting that there's not a test for GoodTuring yet.