This system will allow you to make targeted RAG queries investigating help files. An access code is required in order to distinguish which family of files the user needs to query. To obtain your access code contact Aryan Ghobadi, AI team ([email protected]).
The prerequisites for this system are below, however the startup scripts will install these if your environment is detected to not have them.
- Node.js v16+
- npm
- React
There are some scripts included for your convenience to install prerequisites and deploy the application locally depending on your environment.
If you are a UNIX (MacOS, Linux) user, please run the following command to set up and start the application:
./startup.shIf you are using a Windows environment, run the equivalent startup.bat script to setup and start the application. If Windows Defender prevents this script from running, you can install Node.JS + npm manually and run the setup command in a terminal to setup and deploy the application.
- Node.js and npm: Download and install from Node.js.
-
Open a terminal/command prompt and navigate to the project directory.
-
Run the setup command:
npm run setup
This application uses Node.JS to interact with the Certara AI infrastructure ('Layar') on the backend, answering questions pertinent to a client's specific knowledge base/document family. Your access code is critical in order to ensure that questions asked utilise the correct set of documents when generating responses, ensure this field is filled before sending queries. TL;Dr - To make sure you're asking questions of the right documents, enter your client-specific access code that was provided to you.
This application runs using a Llama 3.1-70b model, which has a context window limit of 128k tokens, for larger document(s), this means that answers will be generated based on the most relevant derived sections up to this limit. As a result, this means that generated answers should be subject to scrutiny and feedback. This feedback will aid us in optimising query parameters/document organisation, maximising accurate retrieval and generation. TL;Dr - Let us know if an answer isn't what it ought to be