This is the data service application for Lingucidity, which is deployed here: https://jk249.azurewebsites.net/
Here is a description of the public endpoints:
| HTTP Method | URL | Description |
|---|---|---|
| GET | / |
a hello world message |
| GET | /user |
authenticate a user, and return credentials if valid |
| GET | /user/{user_id}/phrase |
a list of phrases of a user with given user_id |
| GET | /user/{user_id}/phrase/{phrase_id} |
a phrase with given phrase_id of a user with given user_id |
| POST | /user |
create a new user |
The data service is a RESTful API implemented using Hapi and deployed on Microsoft Azure Web App Service. It provides access to the Lingucidity database hosted on ElephantSQL.
The data service also includes a privately hosted inference endpoint for Meta's Llama 2, an open source 7 billion parameter LLM. The LLM has been quantized to 4-bit precision using AutoAWQ for our purposes and it is downloadable from Calvin University's Hugging Face repository. Please note that this endpoint is not publicly available, and will be removed from the data service in the future.
---
title: Data Service Architecture
---
graph TD;
client[Client] -->|request| data[RESTful API];
client -->|request| llama[LLM];
llama -->|response| client;
data -->|pg-promise| db[Database];
db -->|SQL| data;
data -->|response| client;