-- kepp rasa-x out_of_requirements file rasa-x==0.22.1
sudo apt-get update -y
sudo apt-get install docker.io
sudo curl -L https://github.com/docker/compose/releases/download/1.21.2/docker-compose-`uname -s-uname -m` -o /usr/local/bin/docker-compose
sudo chmod +x /usr/local/bin/docker-compose
cd base_docker;
bash ./base_docker.sh
cd ..
bash ./run.sh
-- done for parsers
add doc-strigns for:
-
parsers
-
actions
-
api
-
api helper
-
endpoints
-
spell checker module
-
Format (standard Python docString format): """ Parse 'domain.yml' file to json object and write to file :param path_to_domain_yml: :param path_to_domain_json: :return: """
-
General Layout: project_root/ │ ├── project/ # Project source code ├── docs/ ├── README ├── HOW_TO_CONTRIBUTE ├── CODE_OF_CONDUCT ├── examples.py
-
example.py: A Python script file that gives simple examples of how to use the projects.
- compiler for Domain Parser (json_domain to yml_domain)
- make classes for all parsers (intent_parser, sotyr_parser, domain_parser)
- dockerize the whole solution
- check for required ports & endpoints
- need to refine code
- code for generating delbrate misspells need to be ingrated with current InvertedIndex generator function in this module 'generate_inverted_index.generate_misspells'
- add in API pipeline
""" Done -> intent parser -> stories parser --> documentation, user manual, api reference --> UI iteration will only be done through json format --> domain parser 1) parse domain to json formats first 2): - dialogflow single click train/test model - domain file dependencies (intents, actions, utter_actions, entities, synonyms) - validation of above files?? UI/Backend?? - source for domain.yml dependencies (==> resources folder in agent) -- entities, synonyms, lookup tables**, utter_actions -- entities and synonyms in a single json file -- actions/utter_hello_user.json ==> separate file for eachlist-of-all-utter-action --> separate model testing for nlu & core - testing through code rather than terminal commands, results manipulation would be easy
- transcribing the latent space of POS
--> train/test split for NLU data & dialogue management
rasa agent dependencies:
- nlu.md (or .md files for each intent)
- stories.md
- domain.yml
- config.yml
- endpoints.yml
test dependencies: (separately is preffered) - test_data for nlu - test data for dialogue management - including all above....
scripts and DB created for intent wise analytics - calling above functions and passing out relent parameters(intent name, )
something to create markdown files create stories with drag & drop - .md to json to flowchat - flowchart to json to .md
? does entity chagnes the accuracy score """
Good developers understand the types of documentation and their importance and that the importance and role varies from type to type.
1- User manuals. This is the how-to software to which users turn when they're figuring things out. How do you configure the software? How do you convert a file? Can you add images? How do you fix a mistake? Even if the solution is a single click on the toolbar or the menu, there may be a lot of potential clicks through which to wade. Good user documentation makes it easier.
2- Project documentation. Details of the project's development are valuable to your team as they work on it and possibly to users who want to customize an open-source program, for instance. The documentation can include contribution policies, best practices, change logs, product requirements, design guidelines and road maps.
3- Requirements documentation. You'll usually draw this up at the start of the project. It defines the expectations for the software, including functional requirements, hardware requirements, compatibility and limitations.
4- Architecture documentation. Defines the high-level architecture of the system: the components, their functions and the data and control flow.
5- Technical documentation. Written for a technical audience, this covers the code, algorithms and interface.