NeuralWOZ: Learning to Collect Task-Oriented Dialogue via Model-based Simulation (ACL-IJCNLP 2021)

Overview

NeuralWOZ

This code is official implementation of "NeuralWOZ: Learning to Collect Task-Oriented Dialogue via Model-based Simulation".

Sungdong Kim, Minsuk Chang, Sang-woo Lee
In ACL 2021.

Citation

@inproceedings{kim2021neuralwoz,
  title={NeuralWOZ: Learning to Collect Task-Oriented Dialogue via Model-based Simulation},
  author={Kim, Sungdong and Chang, Minsuk and Lee, Sang-woo},
  booktitle={ACL},
  year={2021}
}

Requirements

python3.6
torch==1.4.0
transformers==2.11.0

Please install apex for the mixed precision training.
See details in requirements.txt

Data Download and Preprocessing

1. Download dataset

Please run this script at first. It will create data repository, and save and preprocess MultiWOZ 2.1 dataset.

python3 create_data.py

2. Preprocessing

To train NeuralWOZ under various settings, you should create each training instances with running below script.

python3 neuralwoz/preprocess.py --exceptd $TARGET_DOMAIN --fewshot_ratio $FEWSHOT_RATIO
  • exceptd: Specify "target domain" to exclude from training dataset for leave-one-out scheme. It is one of the (hotel|restaurant|attraction|train|taxi).
  • fewshot_ratio: Choose proportion of examples in the target domain to include. Default is 0. which means zero-shot. It is one of the (0.|0.01|0.05|0.1). You can check the fewshot examples in the assets/fewshot_key.json.

This script will create "$TARGET_DOMAIN_$FEWSHOT_RATIO_collector_(train|dev).json" and "$TARGET_DOMAIN_$FEWSHOT_RATIO_labeler_train.h5".

Training NeuralWOZ

You should specify output_path to save the trained model.
Each output consists of the below four files after the training.

  • pytorch_model.bin
  • config.json
  • vocab.json
  • merges.txt

For each zero/few-shot settings, you should set the TRAIN_DATA and DEV_DATA from the preprocessing. For example, hotel_0.0_collector_(train|dev).json should be used for the Collector training when the target domain is hotel in the zero-shot domain transfer task.

We use N_GPU=4 and N_ACCUM=2 for Collector training and N_GPU=2 and N_ACCUM=2 for Labeler training to fit 32 for batch size based on V100 32GB GPU.

1. Collector

python3 neuralwoz/train_collector.py \
  --dataset_dir data \
  --output_path $OUTPUT_PATH \
  --model_name_or_path facebook/bart-large \
  --train_data $TRAIN_DATA \
  --dev_data $DEV_DATA \
  --n_gpu $N_GPU \
  --per_gpu_train_batch_size 4 \
  --num_train_epochs 30 \
  --learning_rate 1e-5 \
  --gradient_accumulation_steps $N_ACCUM \
  --warmup_steps 1000 \
  --fp16

2. Labeler

python3 neuralwoz/train_labeler.py \
  --dataset_dir data \
  --output_path $OUTPUT_PATH \
  --model_name_or_path roberta-base-dream \
  --train_data $TRAIN_DATA \
  --dev_data labeler_dev_data.json \
  --n_gpu $N_GPU \
  --per_gpu_train_batch_size 8 \
  --num_train_epochs 10 \
  --learning_rate 1e-5 \
  --gradient_accumulation_steps $N_ACCUM \
  --warmup_steps 1000 \
  --beta 5. \
  --fp16

Download Synthetic Dialogues from NeuralWOZ

Please download synthetic dialogues from here

  • The naming convention is nwoz_{target_domain}_{fewshot_proportion}.json
  • Each dataset contains synthesized dialogues from our NeuralWOZ
  • Specifically, It contains synthetic dialogues for the target_domain while excluding original dialogues for the target domain (leave-one-out setup)
  • You can check the i-th synthesized dialogue in each files with aug_{target_domain}_{fewshot_proprotion}_{i} for dialogue_idx key.
  • You can use the json file to directly train zero/few-shot learner for DST task
  • Please see readme for training TRADE and readme for training SUMBT using the dataset
  • If you want to synthesize your own dialogues, please see below sections.

Download Pretrained Models

Pretrained models are available in this link. The naming convention is like below

  • NEURALWOZ: (Collector|Labeler)_{target_domain}_{fewshot_proportion}.tar.gz
  • TRADE: nwoz_TRADE_{target_domain}_{fewshot_proportion}.tar.gz
  • SUMBT: nwoz_SUMBT_{target_domain}_{fewshot_proportion}.tar.gz

To synthesize your own dialogues, please download and unzip both of Collector and Labeler in same target domain and fewshot_proportion at $COLLECTOR_PATH and $LABELER_PATH, repectively.

Please use tar -zxvf MODEL.tar.gz for the unzipping.

Generate Synthetic Dialogues using NeuralWOZ

python3 neuralwoz/run_neuralwoz.py \
  --dataset_dir data \
  --output_dir data \
  --output_file_name neuralwoz-output.json \
  --target_data collector_dev_data.json \
  --include_domain $TARGET_DOMAIN \
  --collector_path $COLLECTOR_PATH \
  --labeler_path $LABELER_PATH \
  --num_dialogues $NUM_DIALOGUES \
  --batch_size 16 \
  --num_beams 1 \
  --top_k 0 \
  --top_p 0.98 \
  --temperature 0.9 \
  --include_missing_dontcare

License

Copyright 2021-present NAVER Corp.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
Owner
NAVER AI
Official account of NAVER AI, Korea No.1 Industrial AI Research Group
NAVER AI
An open-source project for applying deep learning to medical scenarios

Auto Vaidya An open source solution for creating end-end web app for employing the power of deep learning in various clinical scenarios like implant d

Smaranjit Ghose 18 May 29, 2022
SAN for Product Attributes Prediction

SAN Heterogeneous Star Graph Attention Network for Product Attributes Prediction This repository contains the official PyTorch implementation for ADVI

Xuejiao Zhao 9 Dec 12, 2022
Tensorflow2 Keras-based Semantic Segmentation Models Implementation

Tensorflow2 Keras-based Semantic Segmentation Models Implementation

Hah Min Lew 1 Feb 08, 2022
VGGFace2-HQ - A high resolution face dataset for face editing purpose

The first open source high resolution dataset for face swapping!!! A high resolution version of VGGFace2 for academic face editing purpose

Naiyuan Liu 232 Dec 29, 2022
Practical and Real-world applications of ML based on the homework of Hung-yi Lee Machine Learning Course 2021

Machine Learning Theory and Application Overview This repository is inspired by the Hung-yi Lee Machine Learning Course 2021. In that course, professo

SilenceJiang 35 Nov 22, 2022
Recall Loss for Semantic Segmentation (This repo implements the paper: Recall Loss for Semantic Segmentation)

Recall Loss for Semantic Segmentation (This repo implements the paper: Recall Loss for Semantic Segmentation) Download Synthia dataset The model uses

32 Sep 21, 2022
Python scripts form performing stereo depth estimation using the CoEx model in ONNX.

ONNX-CoEx-Stereo-Depth-estimation Python scripts form performing stereo depth estimation using the CoEx model in ONNX. Stereo depth estimation on the

Ibai Gorordo 8 Dec 29, 2022
Automatic packaging of the open-composite libs for OvGME

OvGME Packager for OpenXR – OpenComposite for DCS Note This repository is currently unsupported and needs to be migrated to the upstream OpenComposite

12 Nov 03, 2022
💡 Type hints for Numpy

Type hints with dynamic checks for Numpy! (❒) Installation pip install nptyping (❒) Usage (❒) NDArray nptyping.NDArray lets you define the shape and

Ramon Hagenaars 377 Dec 28, 2022
Implementation of various Vision Transformers I found interesting

Implementation of various Vision Transformers I found interesting

Kim Seonghyeon 78 Dec 06, 2022
A deep learning CNN model to identify and classify and check if a person is wearing a mask or not.

Face Mask Detection The Model is designed to check if any human is wearing a mask or not. Dataset Description The Dataset contains a total of 11,792 i

1 Mar 01, 2022
Data pipelines for both TensorFlow and PyTorch!

rapidnlp-datasets Data pipelines for both TensorFlow and PyTorch ! If you want to load public datasets, try: tensorflow/datasets huggingface/datasets

1 Dec 08, 2021
Make your master artistic punk avatar through machine learning world famous paintings.

Master-art-punk Make your master artistic punk avatar through machine learning world famous paintings. 通过机器学习世界名画制作属于你的大师级艺术朋克头像 Nowadays, NFT is beco

Philipjhc 53 Dec 27, 2022
Official PyTorch implementation of "The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person Pose Estimation" (ICCV 21).

CenterGroup This the official implementation of our ICCV 2021 paper The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person P

Dynamic Vision and Learning Group 43 Dec 25, 2022
Simple renderer for use with MuJoCo (>=2.1.2) Python Bindings.

Viewer for MuJoCo in Python Interactive renderer to use with the official Python bindings for MuJoCo. Starting with version 2.1.2, MuJoCo comes with n

Rohan P. Singh 62 Dec 30, 2022
Pytorch implementation of AREL

Status: Archive (code is provided as-is, no updates expected) Agent-Temporal Attention for Reward Redistribution in Episodic Multi-Agent Reinforcement

8 Nov 25, 2022
Price-Prediction-For-a-Dream-Home - A machine learning based linear regression trained model for house price prediction.

Price-Prediction-For-a-Dream-Home ROADMAP TO THIS LINEAR REGRESSION BASED HOUSE PRICE PREDICTION PREDICTION MODEL Import all the dependencies of the p

DIKSHA DESWAL 1 Dec 29, 2021
A pytorch implementation of Detectron. Both training from scratch and inferring directly from pretrained Detectron weights are available.

Use this instead: https://github.com/facebookresearch/maskrcnn-benchmark A Pytorch Implementation of Detectron Example output of e2e_mask_rcnn-R-101-F

Roy 2.8k Dec 29, 2022
TransferNet: Learning Transferrable Knowledge for Semantic Segmentation with Deep Convolutional Neural Network

TransferNet: Learning Transferrable Knowledge for Semantic Segmentation with Deep Convolutional Neural Network Created by Seunghoon Hong, Junhyuk Oh,

42 Jun 29, 2022
A collection of easy-to-use, ready-to-use, interesting deep neural network models

Interesting and reproducible research works should be conserved. This repository wraps a collection of deep neural network models into a simple and un

Aria Ghora Prabono 16 Jun 16, 2022