EasyIntentCatcher

Administrative tool for managing datasets of IntentCacther

Features:

• IntentCatcher dataset management: view, edit, create new training samples from WEB UI
• Train IntentCacther models from UI
• Test and Analyze predictions of the trained model
• Export trained models to remote SSH Server
• Automatically prepares model for updated dataset

TODO add screenshots:

• intents list intent details
• predictions log
• prediction api

Requirements:

Assure your server fit following requirements:

• python 3.7.9
  • conda -N new_env python=3.7
• redis-server
  • sudo apt-get install redis-server

Set Up

How to setup:

git clone https://github.com/deepmipt/EasyIntentCacther
cd EasyIntentCacther
pip install -r requirements.txt
python manage.py migrate

Now you need to launch worker for training tasks and web server for managing dataset and training system.

Run worker for training task execution:

celery  -A EasyIntentCatcher worker -l INFO

Run server for web-administration tool

python manage.py runserver 0.0.0.0:8000

Now you can browse your intents, train a model, export to SSH server and use it for predictions!

Useful commands for work with IntentCatcher on lowel level interface

Generate IntentCather JSON dataset specification from Database contents:

python ic_dataset/from_db_2_icjson.py

Train model from IntentCather JSON dataset specification with specification of target path for model:

python data/create_data_and_train_model.py --intent_phrases_path data/intent_phrases_export.json --model_path data/models/my_model

Screenshots

List of intents:

Intent editing:

History of predictions:

History of trained models:

Prediction API:

API Response example: