Text is no more Enough! A Benchmark for Profile-based Spoken Language Understanding

This repository contains the official PyTorch implementation of the paper:

Text is no more Enough! A Benchmark for Profile-based Spoken Language Understanding. Xiao Xu*, Libo Qin*, Kaiji Chen, Guoxing Wu, Linlin Li, Wanxiang Che. AAAI 2022. [Paper(Arxiv)] [Paper]

If you use any source codes or the datasets included in this toolkit in your work, please cite the following paper. The bibtex are listed below:

...
example

In the following, we will guide you how to use this repository step by step.

Workflow

framework

Architecture

framework

Results

result

Preparation

Our code is based on the following packages:

  • numpy==1.19.5
  • tqdm==4.50.2
  • pytorch==1.7.0
  • python==3.7.3
  • cudatoolkit==11.0.3
  • transformers==4.1.1

We highly suggest you using Anaconda to manage your python environment.

We download the chinese pretrained model checkpoints from the following links:

How to Run it

The script train.py acts as a main function to the project, you can run the experiments by the following commands.

# LSTM w/o Profile on TITAN Xp
python train.py -g -fs -es -uf -bs 8 -lr 0.0006
# LSTM w/ Profile on TITAN Xp
python train.py -g -fs -es -uf -ui -bs 8 -lr 0.0004
# BERT w/o Profile on Tesla V100s PCIE 32GB
python train.py -g -fs -es -uf -up -mt XLNet -bs 8 -lr 0.001 -blr 4e-05
# BERT w/ Profile on Tesla V100 PCIE 32GB
python train.py -g -fs -es -uf -up -ui -mt ELECTRA -bs 8 -lr 0.0008 -blr 4e-05

If you have any question, please issue the project or email me or lbqin, and we will reply you soon.

Acknowledgement

  • We are highly grateful for the public code of Stack-Propagation!

    A Stack-Propagation Framework with Token-Level Intent Detection for Spoken Language Understanding. Libo Qin,Wanxiang Che, Yangming Li, Haoyang Wen and Ting Liu. (EMNLP 2019). Long paper. [pdf] [code]

  • We are highly grateful for the open-source knowledge graph!

GitHub

View Github