Home

Awesome

Slot-Utterance Matching for Universal and Scalable Belief Tracking

This is the original PyTorch implemenation of SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking, Hwaran Lee*, Jinsik Lee*, and Tae-Yoon Kim, ACL 2019 (Short)

Requirements

Usages

Data prepration & pre-procesisng

Train SUMBT

Please see run.sh

python3 code/Main-multislot.py --do_train --do_eval --data_dir data/woz --bert_model bert-base-uncased --do_lower_case --task_name bert-gru-slot_query_multi --nbt rnn --output_dir exp-woz/model --target_slot all 

Experiment results on MultiWOZ

python3 code/main-multislot.py --do_train --do_eval --num_train_epochs 300 --data_dir data/multiwoz --bert_model bert-base-uncased --do_lower_case --task_name bert-gru-sumbt --nbt rnn --output_dir exp-multiwoz/model --target_slot all --warmup_proportion 0.1 --learning_rate 1e-4 --train_batch_size 3 --eval_batch_size 16 --distance_metric euclidean --patience 15 --tf_dir tensorboard --hidden_dim 300 --max_label_length 32 --max_seq_length 64 --max_turn_length 22
Hidden dimJoint acc.Slot acc.Joint acc. (Restaurant)Slot acc. (Restaurant)
3000.488060.973290.828540.96537
6000.490640.972900.828400.96475

Notes and Acknowledgements

The code is developed based on PyTorch BERT from https://github.com/huggingface/pytorch-pretrained-BERT and The Annotated Transformer

Citation

@inproceedings{lee2019sumbt,
  title={SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking},
  author={Lee, Hwaran and Lee, Jinsik and Kim, Tae-Yoon},
  booktitle={Proceedings of the 57th Conference of the Association for Computational Linguistics},
  pages={5478--5483},
  year={2019}
}

Contact Information

Contact: Hwaran Lee (hwaran.lee@gmail.com), Jinsik Lee (jinsik16.lee@sktbrain.com), Tae-Yoon Kim (oceanos@sktbrain.com)