www.icact.org

Member Login

TACT Journal Homepage

Author Homepage

Call for Paper

Paper Procedure

Paper Submission/Update

Invoice/Registration

CRF Submission/Update

Find My Paper

Statistics & History

Welcome Message

Committee

FAQ

Paper Archives

Outstanding Papers

Program & Proceedings

Presentation Platform

Hotel & Travel Info

Photo Gallery

Scheduler Login

Seminar

Archives Login

Sponsors

Virual Oral Question and Answer


IEEE/ICACT20210164 Question.5 Questioner: quandh@vnu.edu.vn 2021-02-09 오후 7:46:47	IEEE/ICACT20210164 Answer.5 Answer by Auhor freemanpolys@gmail.com 2021-02-09 오후 7:46:47
Thanks for your presentation. I'd like to know more about your training data. Did you build it from scratch or use any publicly available one. Besides, can you please detail your parameters in order to customize the model and finetune it on your dataset?	We build it from scratch but note that for the NLU pipeline (tokenizers, featurizers and training algorithms) we use public models : WhitespaceTokenizer, RegexFeaturizer, CountVectorsFeaturizer, DIETClassifier ... For the model, it's the DIETClassifier which is used with epochs value 100. Our configurations is based on the Rasa doc suggested one . You can find them here : https://rasa.com/docs/rasa/model-configuration

Select Voice