Label Studio
A flexible data labeling platform for fine-tuning LLMs, preparing training data, and validating AI models
Overview
Label Studio is an open-source data labeling platform. It is designed to manage multiple projects, users, and data types in a single platform. With this flexibility, you can perform different types of labeling for multiple data types and can integrate it with LLMs to perform pre-labeling and continuous active learning.
This Starter deploys Label Studio with a single click. Once deployed, you can log in with your provided credentials at the Koyeb service's public URL to access the web interface and begin using the platform.
Configuration
Label Studio requires at least 8GB of ram and recommends 16GB.
The following environment variables are available with this image:
LABEL_STUDIO_USERNAME: The username (in email address format) to use when authenticating to Label Studio.LABEL_STUDIO_PASSWORD: The password to use when authenticating to Label Studio.LABEL_STUDIO_HOST: The hostname where Label Studio will run. Set this tohttps://{{ KOYEB_PUBLIC_DOMAIN }}to automatically set the correct value.DATABASE_URL: An optional PostgreSQL connection string if you wish to use an external PostgreSQL database instead of an embedded SQLite database.
Once the service is up and running, you can log in to the service's public URL using the username and password you configured. You may wish to configure external storage using an object storage provider to persist data across deployments.
Other resources related to Label Studio
