Enable auto labeling with OpenAI or MistralAI

To simplify and speed up document labeling, WorkFusion implemented an automatic document labeling mechanism based on OpenAI (ChatGPT) and MistralAI LLM (large language model) services.

You can activate both services simultaneously and select one of them to do the auto labeling of document datasets in the AutoML UI.

Activate auto labeling with OpenAI

To activate OpenAI-based labeling, follow the steps below:

1. Add the OpenAI key to Secrets Vault:

   cd /opt/workfusion/wf-sec-storage
   
   create file `autolabeling.properties` with content:
   
   openai.key=<YOUR OPEN AI KEY>
   
   Run command:
   ./loader.sh automl-worker autolabeling.properties
   
   Check updated property value (openai.key):
   ./loader.sh automl-worker review
   
   You should see something like:
   [INFO ] Alias : Key : Value
   [INFO ] openai.key : openai.key : <YOUR OPEN AI KEY>
   [INFO ] It's all

2. Set the /config/labeling-service/openai.enabled property in ZooKeeper to true. You can use ZooNavigator to do this.

3. Restart the labeling service:

   wfmanager restart labeling-service

Now, you can use OpenAI (ChatGPT) models for automatic labeling of datasets. For details, see the Manage datasets | Labeling with AI guide.

Activate auto labeling with MistralAI

To activate MistralAI-based labeling, follow the steps below:

1. Add the MistralAI key to Secrets Vault:

   cd /opt/workfusion/wf-sec-storage
   
   create file `autolabeling.properties` with content:
   
   mistralai.key=<YOUR MISTRAL AI KEY>
   
   Run command:
   ./loader.sh automl-worker autolabeling.properties
   
   Check updated property value (openai.key):
   ./loader.sh automl-worker review
   
   You should see something like:
   [INFO ] Alias : Key : Value
   [INFO ] mistralai.key : mistralai.key : <YOUR MISTRAL AI KEY>
   [INFO ] It's all

2. Set the /config/labeling-service/mistralai.enabled property in ZooKeeper to true. You can use ZooNavigator to do this.

3. Restart the labeling service:

   wfmanager restart labeling-service

Now, you can use MistralAI models for the automatic labeling of datasets. For details, see the Manage datasets | Labeling with AI guide.