flair

Flair NER example

This example demonstrates how to use Flair NER model with Label Studio.

Before you begin

Before you begin, you must install the Label Studio ML backend.

This tutorial uses the flair example.

Quickstart

1. Build and start the Machine Learning backend on http://localhost:9090

docker-compose up

2. Validate that the backend is running

$ curl http://localhost:9090/health
{"status":"UP"}

3. Create a project in Label Studio. Then from the Model page in the project settings, connect the model. The default URL is http://localhost:9090.

Labeling Configuration

<View>
  <Labels name="label" toName="text">
    <Label value="PER" background="red"/>
    <Label value="ORG" background="darkorange"/>
    <Label value="LOC" background="orange"/>
    <Label value="MISC" background="green"/>
  </Labels>

  <Text name="text" value="$text"/>
</View>

Parameters

• FLAIR_MODEL_NAME: The name of the Flair model to use. Default is ner. See all options here