This work is published as "A Sequence-to-Sequence Approach to Dialogue State Tracking" (ACL 2021)
Three example scripts are provided for how to use Seq2Seq-DU. train.sh will train a simpler model, test.sh will predict dialogue state on the test dataset, and eval.sh will evaluate the performance.
1.absl-py (for tests)
2.fuzzywuzzy
3.numpy
4.six
5.tensorflow