Source code for our paper "Empathetic Response Generation with State Management"

Overview

Source code for our paper "Empathetic Response Generation with State Management"

this repository is maintained by both Jun Gao and Yuhan Liu

Model Overview

model

Environment Requirement

  • pytorch >= 1.4
  • sklearn
  • nltk
  • numpy
  • bert-score

Dataset

you can directly use the processed dataset located in data/empathetic:

├── data
│   ├── empathetic
│   │   ├── parsed_emotion_Ekman_intent_test.json
│   │   ├── parsed_emotion_Ekman_intent_train.json
│   │   ├── parsed_emotion_Ekman_intent_valid.json
│   │   ├── emotion_intent_trans.mat
│   │   ├── goEmotion_emotion_trans.mat

Or you want to reproduce the data annotated with goEmotion emotion classifier and empathetic intent classifier, you can run the command:

  • convert raw csv empathetic dialogue data into json format. (origin dataset link: EmpatheticDialogues)

    bash preprocess_raw.sh
  • train emotion classfier with goEmotion dataset and annotate (origin dataset link: goEmotion). Here $BERT_DIR is your pretrained BERT model directory which includes vocab.txt, config.json and pytorch_model.bin, here we simply use bert-base-en from Hugginface

    bash ./bash/emotion_annotate.sh  $BERT_DIR 32 0.00005 16 3 1024 2 0.1
  • train intent classfier with empathetic intent dataset and annotate (origin dataset link: Empathetic_Intent)

    bash ./bash/intent_annotate.sh  $BERT_DIR 32 0.00005 16 3 1024 2 0.1
  • build prior emotion-emotion and emotion-intent transition matrix

    bash ./bash/build_transition_mat.sh

Train

For training the LM-based model, you need to download bert-base-en and gpt2-small from Hugginface first, then run the following command. Here $GPT_DIR and $BERT_DIR are the downloaded model directory:

bash ./bash/train_LM.sh --gpt_path $GPT_DIR --bert_path $BERT_DIR --gpu_id 2 --epoch 5 --lr_NLU 0.00003 --lr_NLG 0.00008 --bsz_NLU 16 --bsz_NLG 16

for example:

bash ./bash/train_LM.sh --gpt_path /home/liuyuhan/datasets/gpt2-small --bert_path /home/liuyuhan/datasets/bert-base-en bert-base-en --gpu_id 2 --epoch 5 --lr_NLU 0.00003 --lr_NLG 0.00008 --bsz_NLU 16 --bsz_NLG 16

For training the Trs-based model, we use glove.6B.300d as the pretrained word embeddings. You can run the following command to train model. Here $GLOVE is the glove embedding txt file.

bash ./bash/train_Trs.sh --gpu_id 2 --epoch 15 --lr_NLU 0.00007 --lr_NLG 0.0015 --bsz_NLU 16 --bsz_NLG 16 --glove $GLOVE

for example:

bash ./bash/train_Trs.sh --gpu_id 2 --epoch 15 --lr_NLU 0.00007 --lr_NLG 0.0015 --bsz_NLU 16 --bsz_NLG 16 --glove /home/liuyuhan/datasets/glove/glove.6B.300d.txt

Evaluate

To generate the automatic metric results, firstly you need to make sure that bert-score is successfully installed. In our paper, we use roberta-large-en rescaled with baseline to calculate BERTScore. You can download roberta-large-en from Hugginface. For the rescaled_baseline file, we can download it from here and put it under the roberta-large-en model directory.

Then you can run the following command to get the result, here $hypothesis and $reference are the generated response file and ground-truth response file. $result is the output result file. $ROBERTA_DIR is the downloaded roberta-large-en model directory.

To evaluate LM-based model, the command is:

bash ./bash/eval.sh --hyp $hypothesis --ref ./data/empathetic/ref.txt --out $result --bert $ROBERTA_DIR --gpu_id 0 --mode LM

To evaluate Trs-based model, the command is:

bash ./bash/eval.sh --hyp $hypothesis --ref ./data/empathetic/ref_tokenize.txt --out $result --bert $ROBERTA_DIR --gpu_id 0 --mode Trs
Owner
Yuhan Liu
NLPer
Yuhan Liu
mmdetection version of TinyBenchmark.

introduction This project is an mmdetection version of TinyBenchmark. TODO list: add TinyPerson dataset and evaluation add crop and merge for image du

34 Aug 27, 2022
[ICCV 2021] Official Tensorflow Implementation for "Single Image Defocus Deblurring Using Kernel-Sharing Parallel Atrous Convolutions"

KPAC: Kernel-Sharing Parallel Atrous Convolutional block This repository contains the official Tensorflow implementation of the following paper: Singl

Hyeongseok Son 50 Dec 29, 2022
PyTorch implementation of "Contrast to Divide: self-supervised pre-training for learning with noisy labels"

Contrast to Divide: self-supervised pre-training for learning with noisy labels This is an official implementation of "Contrast to Divide: self-superv

55 Nov 23, 2022
Multi-Task Learning as a Bargaining Game

Nash-MTL Official implementation of "Multi-Task Learning as a Bargaining Game". Setup environment conda create -n nashmtl python=3.9.7 conda activate

Aviv Navon 87 Dec 26, 2022
The official implementation of CircleNet: Anchor-free Detection with Circle Representation, MICCAI 2030

CircleNet: Anchor-free Detection with Circle Representation The official implementation of CircleNet, MICCAI 2020 [PyTorch] [project page] [MICCAI pap

The Biomedical Data Representation and Learning Lab 45 Nov 18, 2022
Official code repository for A Simple Long-Tailed Rocognition Baseline via Vision-Language Model.

BALLAD This is the official code repository for A Simple Long-Tailed Rocognition Baseline via Vision-Language Model. Requirements Python3 Pytorch(1.7.

peng gao 42 Nov 26, 2022
NeWT: Natural World Tasks

NeWT: Natural World Tasks This repository contains resources for working with the NeWT dataset. ❗ At this time the binary tasks are not publicly avail

Visipedia 26 Oct 18, 2022
Reproduce ResNet-v2(Identity Mappings in Deep Residual Networks) with MXNet

Reproduce ResNet-v2 using MXNet Requirements Install MXNet on a machine with CUDA GPU, and it's better also installed with cuDNN v5 Please fix the ran

Wei Wu 531 Dec 04, 2022
Transfer Reinforcement Learning for Differing Action Spaces via Q-Network Representations

Transfer-Learning-in-Reinforcement-Learning Transfer Reinforcement Learning for Differing Action Spaces via Q-Network Representations Final Report Tra

Trung Hieu Tran 4 Oct 17, 2022
Dirty Pixels: Towards End-to-End Image Processing and Perception

Dirty Pixels: Towards End-to-End Image Processing and Perception This repository contains the code for the paper Dirty Pixels: Towards End-to-End Imag

50 Nov 18, 2022
Utilizes Pose Estimation to offer sprinters cues based on an image of their running form.

Running-Form-Correction Utilizes Pose Estimation to offer sprinters cues based on an image of their running form. How to Run Dependencies You will nee

3 Nov 08, 2022
A PyTorch Implementation of ViT (Vision Transformer)

ViT - Vision Transformer This is an implementation of ViT - Vision Transformer by Google Research Team through the paper "An Image is Worth 16x16 Word

Quan Nguyen 7 May 11, 2022
Individual Treatment Effect Estimation

CAPE Individual Treatment Effect Estimation Run CAPE python train_causal.py --loop 10 -m cape_cau -d NI --i_t 1 Run a baseline model python train_cau

S. Deng 4 Sep 02, 2022
ONNX-GLPDepth - Python scripts for performing monocular depth estimation using the GLPDepth model in ONNX

ONNX-GLPDepth - Python scripts for performing monocular depth estimation using the GLPDepth model in ONNX

Ibai Gorordo 18 Nov 06, 2022
Contrastive Language-Image Pretraining

CLIP [Blog] [Paper] [Model Card] [Colab] CLIP (Contrastive Language-Image Pre-Training) is a neural network trained on a variety of (image, text) pair

OpenAI 11.5k Jan 08, 2023
A concise but complete implementation of CLIP with various experimental improvements from recent papers

x-clip (wip) A concise but complete implementation of CLIP with various experimental improvements from recent papers Install $ pip install x-clip Usag

Phil Wang 515 Dec 26, 2022
YOLOv2 in PyTorch

YOLOv2 in PyTorch NOTE: This project is no longer maintained and may not compatible with the newest pytorch (after 0.4.0). This is a PyTorch implement

Long Chen 1.5k Jan 02, 2023
Python wrapper of LSODA (solving ODEs) which can be called from within numba functions.

numbalsoda numbalsoda is a python wrapper to the LSODA method in ODEPACK, which is for solving ordinary differential equation initial value problems.

Nick Wogan 52 Jan 09, 2023
Pytorch implementation of Integrating Tree Path in Transformer for Code Representation

This is an official Pytorch implementation of the approaches proposed in: Han Peng, Ge Li, Wenhan Wang, Yunfei Zhao, Zhi Jin “Integrating Tree Path in

Han Peng 16 Dec 23, 2022
Council-GAN - Implementation for our paper Breaking the Cycle - Colleagues are all you need (CVPR 2020)

Council-GAN Implementation of our paper Breaking the Cycle - Colleagues are all you need (CVPR 2020) Paper Ori Nizan , Ayellet Tal, Breaking the Cycle

ori nizan 260 Nov 16, 2022