GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → ishalyminov → few_shot_dialogue_generation

ishalyminov / few_shot_dialogue_generation

Licence: other
Dialogue Knowledge Transfer Networks (DiKTNet)

Programming Languages

python
139335 projects - #7 most used programming language
Jupyter Notebook
11667 projects
shell
77523 projects

Labels

nlpmachine-learningnatural-language-processingdeep-learningdialogpytorchtransfer-learningemnlpdialogue-systemszero-shot-learningdialogue-generationconversational-aifew-shot-learningemnlp2019sigdialsigdial2019diktnet

Projects that are alternatives of or similar to few shot dialogue generation

Awesome Domain Adaptation
A collection of AWESOME things about domian adaptation
Stars: ✭ 3,357 (+13887.5%)
Mutual labels:  transfer-learning, zero-shot-learning, few-shot-learning
adapt
Awesome Domain Adaptation Python Toolbox
Stars: ✭ 46 (+91.67%)
Mutual labels:  transfer-learning, zero-shot-learning, few-shot-learning
Transferlearning
Transfer learning / domain adaptation / domain generalization / multi-task learning etc. Papers, codes, datasets, applications, tutorials.-迁移学习
Stars: ✭ 8,481 (+35237.5%)
Mutual labels:  transfer-learning, few-shot-learning
permuted-bAbI-dialog-tasks
Dataset for 'Learning End-to-End Goal-Oriented Dialog with Multiple Answers' EMNLP 2018
Stars: ✭ 17 (-29.17%)
Mutual labels:  dialog, dialogue-systems
ContrastiveLearning4Dialogue
The codebase for "Group-wise Contrastive Learning for Neural Dialogue Generation" (Cai et al., Findings of EMNLP 2020)
Stars: ✭ 54 (+125%)
Mutual labels:  dialog, conversational-ai
finetuner
Finetuning any DNN for better embedding on neural search tasks
Stars: ✭ 442 (+1741.67%)
Mutual labels:  transfer-learning, few-shot-learning
Meta-Fine-Tuning
[CVPR 2020 VL3] The repository for meta fine-tuning in cross-domain few-shot learning.
Stars: ✭ 29 (+20.83%)
Mutual labels:  transfer-learning, few-shot-learning
Talkit
Non-Linear Game Dialogue Editor
Stars: ✭ 83 (+245.83%)
Mutual labels:  dialog, dialogue-systems
Gonorth
GoNorth is a story and content planning tool for RPGs and other open world games.
Stars: ✭ 289 (+1104.17%)
Mutual labels:  dialog, dialogue-systems
Rakugo Archive
Framework (inspired by Ren'Py) for story driven games in Godot.
Stars: ✭ 291 (+1112.5%)
Mutual labels:  dialog, dialogue-systems
Transfer Learning Conv Ai
🦄 State-of-the-Art Conversational AI with Transfer Learning
Stars: ✭ 1,217 (+4970.83%)
Mutual labels:  dialog, transfer-learning
CVAE Dial
CVAE_XGate model in paper "Xu, Dusek, Konstas, Rieser. Better Conversations by Modeling, Filtering, and Optimizing for Coherence and Diversity"
Stars: ✭ 16 (-33.33%)
Mutual labels:  emnlp, dialogue-generation
HINT3
This repository contains datasets and code for the paper "HINT3: Raising the bar for Intent Detection in the Wild" accepted at EMNLP-2020's Insights Workshop https://insights-workshop.github.io/ Preprint for the paper is available here https://arxiv.org/abs/2009.13833
Stars: ✭ 27 (+12.5%)
Mutual labels:  dialogue-systems, conversational-ai
Dialogue
Node based dialogue system
Stars: ✭ 207 (+762.5%)
Mutual labels:  dialog, dialogue-systems
explicit memory tracker
[ACL 2020] Explicit Memory Tracker with Coarse-to-Fine Reasoning for Conversational Machine Reading
Stars: ✭ 35 (+45.83%)
Mutual labels:  dialogue-systems, conversational-ai
DialoFlow
Code for ACL 2021 main conference paper "Conversations Are Not Flat: Modeling the Dynamic Information Flow across Dialogue Utterances".
Stars: ✭ 85 (+254.17%)
Mutual labels:  dialogue-systems, dialogue-generation
Wisty.js
🧚‍♀️ Chatbot library turning conversations into actions, locally, in the browser.
Stars: ✭ 24 (+0%)
Mutual labels:  dialogue-systems, conversational-ai
RECCON
This repository contains the dataset and the PyTorch implementations of the models from the paper Recognizing Emotion Cause in Conversations.
Stars: ✭ 126 (+425%)
Mutual labels:  dialogue-systems, dialogue-generation
dialogue-datasets
collect the open dialog corpus and some useful data processing utils.
Stars: ✭ 24 (+0%)
Mutual labels:  dialog, dialogue-systems
Unit Uskit
unit-uskit
Stars: ✭ 197 (+720.83%)
Mutual labels:  dialog, dialogue-systems
View All Similar Projects ➔

Dialogue Knowledge Transfer Networks (DiKTNet)

This codebase is shared by the following papers: "Few-Shot Dialogue Generation Without Annotated Data: A Transfer Learning Approach" by Igor Shalyminov, Sungjin Lee, Arash Eshghi, and Oliver Lemon. [SigDial 2019 publication] [bibtex] [Poster]

"Data-Efficient Goal-Oriented Conversation with Dialogue Knowledge Transfer Networks" by Igor Shalyminov, Sungjin Lee, Arash Eshghi, and Oliver Lemon. [EMNLP 2019 publication] [bibtex] [Poster]

If you find it useful for your work, please cite the papers above.

The versions of the code corresponding to each publication can be found using the git release tags.

Pre-requisites

  1. Download the Maluuba MetaLWOz (previously called BLIS) dataset (assuming you extract it into metalwoz-v1 folder)
  2. Prepare environment and dependencies. Below are the steps for Conda:
  conda create -n diktnet python=3.6
  conda activate diktnet
  git submodule update --init
  pip install -r requirements.txt
  1. Create a vocabulary for LAED training: python make_vocabulary.py <data-dir> <corpus client type> <vocab_file.json>

Given the data sources we have, it may be NeuralDialog-ZSDG/data/stanford/ZslStanfordCorpus, NeuralDialog-ZSDG/data/simdial/SimDialCorpus or metalwoz-v1/LAEDBlisCorpus. Vocabularies of a higher coverage can be produced by merging the primary ones.

How To

  1. Training a LAED model - StED (skip-thought dialog-level)
python st_ed.py \
    metalwoz-v1 \
    LAEDBlisCorpus \
    vocabs/maluuba.json \
   --exlcude_domains ... \
   --y_size 10

1.1 Training a LAED model - vanilla VAE (dialog-level)

python vae.py \
  metalwoz-v1 \
  LAEDBlisCorpus \
  vocabs/maluuba.json \
 --exlcude_domains ... \
 --y_size 1
 --k 100
  1. Generating LAED features - StED (skip-thought dialog-level)
python generate_laed_features.py \
    laed_models/st_ed_maluuba_${TARGET_DOMAIN} \
    laed_features/st_ed_maluuba__smd_${TARGET_DOMAIN} \
    --model_name StED \
    --model_type dialog \
    --data_dir NeuralDialog-ZSDG/data/stanford \
    --corpus_client ZslStanfordCorpus \
    --data_loader SMDDialogSkipLoader \
    --vocab maluuba.json
  1. Training a Few-Shot Dialog Generation model
python train_fsdg.py \
    LAZslStanfordCorpus \
    --data_dir NeuralDialog-ZSDG/data/stanford \
    --laed_z_folders laed_features/st_ed_3x5_maluuba__smd_${domain} \
    --black_domains $domain \
    --black_ratio 0.9 \
    --action_match False \
    --target_example_cnt 0 \
    --random_seed $rnd
Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].