AI API

This is a simple API that allows you to use a combination of different methods to deliver GPT text generation.

Installation

git clone https://github.com/beauthebeau/ai-api.git
pip install -r requirements.txt
# Also see https://pytorch.org/get-started/locally/

Usage

python main.py # Starts the Uvicorn server

Endpoints

What is returned

All endpoints return the following:

{
    "text": "The generated text"
}

v1 routes

GET /generate/{prompt}

Generates text based on the prompt provided. No context is provided, so the model will generate text based on the prompt only. Returns a JSON object with the generated text.

GET /generate/{prompt}/{context}

Generates text based on the prompt and context provided. The context is provided as a string, and the model will generate text based on the prompt and context. Returns a JSON object with the generated text.

v2 routes

Uses DialoGPT, by Microsoft and trained on Reddit data to generate text. These routes use HuggingFace Transformers.

GET /generate/v2/?prompt={prompt}

Generates text based on the prompt provided. No context is provided, so the model will generate text based on the prompt only.

v3 routes

Uses OneAI's GPT Skill to generate text. The default model is text-davinci-003. Note that this is a legacy model, superseded by gpt-3.5-turbo

GET /generate/v3/?prompt={prompt}