multiwoz

multiwoz copied to clipboard

[ 

Gs-Zhang

Randomness in evaluation?

2

I spotted some randomness in evaluation code. For example, https://github.com/budzianowski/multiwoz/blob/e4922d6a35d454c1c8152bbf93d476633461606a/model/evaluator.py#L142 Wouldn't it make the match and success rates different even if we evaluate the same model twice?

shaform

generate SLU datasets

1

Does it possible to generate the data set of the SLU task, that is, the data set of the sequence labeling task, I need to obtain all possible values of...

niartnelis

Hello, we have proposed a framework(CUDA-dst) that can do data augmentation on MultiWOZ in DST. The best JGA score we have achieved is 62.93% for TripPy + CUDA(without coreference). Arxiv:...

Mecoli1219

Evaluation Dataset in "Response Generation" - "end-to-end models" part

1

Is this part used multiwoz 2.2 or multiwoz 2.0 as a benchmark dataset. I'm so confused, in RewardNet, Mars, KRLS original paper, all results are the same as your table,...

nqchieutb01

WOZ instructions aka "message" in MultiWOZ2.2

1

Hi! Where I can find the "message" field from the MultiWOZ2.1 format in the MultiWOZ2.2. Ie. where I can find instructions for crowdsource workers in MultiWOZ2.2 format? This is a...

oplatek

Active Intent Labeling Issues

1

I am trying to understand the role that active intents play in the dataset and replicate Active Intent Accuracy (MultiWoz 2.1) 0.924 [from](https://github.com/google-research/google-research/tree/master/schema_guided_dst#results) . But I've noticed a lot of...

KeremTurgutlu

Hi, In **multiwoz**, inappropriate dependency versioning constraints can cause risks. Below are the dependencies and version constraints that the project is using ``` torch numpy nltk==3.4.5 scipy tqdm ``` The...

PyDeps

Hi there, Is there an easy way to convert the MultiWoZ 2.1 dataset to the same format as the 2.2? Best, Pietro

pietrolesci

I saw that this dataset was used for dialogue summarization in some papers. But I can't find "summary" column. How and Where can I get "summary" data for each "dialogue"?

SangwonPark0211