Normalization during test

[gusalsdmlwlq]

Hi.

multiwoz/evaluate.py

Line 133 in a24d299

val2 = normalize(val2)

multiwoz/utils/nlp.py

Line 73 in a24d299

text = re.sub(timepat, ' [value_time] ', text)

In this line, the values of belief state are normalized before searching DB.

But, the time values of train(leaveAt, arrvieBy) are also normalized to [value_time].

In this case, I think we cannot check whether the values meet the user's goal, right?

Is it intended?

[budzianowski]

Hi, thanks for catching this! It's a bug!
However, due to backwards compatibility with previous results, it won't be easy to fix that straight away.
I will try to come up with a good solution.

[gusalsdmlwlq]

Thanks for relpy!
Well, I'll leave this issue open.
If you want or resolve, close this issue please.