Cosmian AI Backend

Confidential computing backend to run language models

Usage

• Build and install the app

• Edit the config file (more info)

• Run the app

CONFIG_PATH="./run/config.json" cosmian-ai-runner --port 5001

• Create a supervisord service:

Sample config file: ./run/cosmian-ai-backend.conf

API endpoints

/summarize

Request:

• doc: content of the document to summarize (String),

• src_lang: optional source language of the text to summarize (String)

Response:

• summary: result (String)

/translate

Request:

• doc: content of the document to translate (String),

• src_lang: source language of the text to translate (String),

• tgt_lang: desired output language (String)

Response:

• translation: result (String)