LakeGPT Project

Architecture

• Components
  • Pytorch
  • HuggingFace Transformers ("seeklhy/codes-1b")
  • Chroma DB as Vector DB
  • FastAPI

REST API Endpoints

Authentication

• POST /token
  • Authenticate user and get access token
  • Request Body: username, password (form-data)
  • Response: JWT access token

Core Endpoints

• GET /
  • Root endpoint
  • Requires: Authorization header
  • Returns: Welcome message

Question Processing

• POST /process_question
  • Process natural language questions
  • Requires: Authorization header
  • Request Body:

    {
      "question": "string"
    }
    

  • Returns: Generated SQL query

Table Management

• GET /tables
  • List all available tables
  • Requires: Authorization header
  • Returns: Array of table names

Question History

• POST /save_question

  • Save a question for future reference
  • Requires: Authorization header
  • Request Body:

    {
      "question": "string"
    }
    

• GET /questions

  • List all saved questions
  • Requires: Authorization header
  • Returns: Array of saved questions

Authentication

All endpoints except /token require Bearer token authentication:

Authorization: Bearer <your_access_token>

Getting Started

To apply the configmap, run:

kubectl apply -f configmap.yaml