OLIM

Open Labeller for Iterative Machine learning

OLIM is a simple labelling interface aimed to be used by personal without data science specific knowledge. Currently OLIM supports text labelling. In future versions we expect to expand the possibilities of the data shape.

Installation

1. Clone the repository:

   git clone https://gitlab.com/nanogennari/olim.git

2. Enter the directory:

   cd olim

3. Configure environment:

   cp .env.template .env

   Edit .env file according to your needs.

4. Build and start the containers:

   docker compose up -d

5. Go to http://localhost:42000 and complete the initial setup via the web interface.

Distributed Workers Deployment

For scalable deployments, you can separate workers (previously called learner) from the main application. Important: Start workers first, then the main application.

Step 0: Common Configuration

Generate a common .env file as in the normal setup above. Use this as the base configuration for both main server and worker machines.

Step 1: Configure and Start Workers First

On workers machine, edit .env to change only these variables from the common configuration:

# Worker-specific changes to .env:
DB_HOST=192.168.1.100  # IP of main server
ES_SERVER=http://192.168.1.100:9200  # Elasticsearch on main server

Start workers:

# On worker machines (start these FIRST)
docker-compose -f docker-compose.workers-only.yml up -d

Step 2: Configure and Start Main Application

On main server, edit .env to change only this variable from the common configuration:

# Main server-specific changes to .env:
REDIS_HOST=192.168.1.200  # IP of worker machine

Start main application:

# On main server (start AFTER workers are running)
docker-compose -f docker-compose.remote-workers.yml up -d