A http server for exposing a gpt4All API
- node version >= 18.0.0
- npm version >= 10.1.0
npm install
npm start
- install requirements
- install and run the node commands above
- modify constants if needed
- first run it will download the model specified in MODEL_NAME constant file to the MODEL_PATH directory
- if you want to disable personas behavior and make it run faster uncomment line in gptCompletion.js
// systemPromptTemplate: '### System:'
{
"prompt": "what is water made of?",
"persona": "default",
"flush": true
}
- prompt: is the text to be completed
- persona: is the persona to be used
- the src/personas directory contains the personas to be used
- a persona is a text file with the persona prompt configuration
- flush: is an optional boolean to flush the historical conversations for the persona, src/history directory contains the historical conversations for each persona