Efficient Data Annotation With Annolive’s Bulk Annotation

Welcome to this guide to bulk annotate your whole data with a single click in Annolive, using GPT3.5 or your own trained model. In the fast paced AI world, it is important to get the data ready quickly. Bulk Annotation feature in Annolive lets you to label the data using LLM/pretrained models in a single click. Users can further refine the bulk annotated result for better accuracy. This in turn reduce the effort on the annotation task by 80%

Bulk Annotation workflow

What are we going to do:

  1. Select the Task in the project
  2. Configure the models for auto-annotation
  3. Bulk Annotate and refine the data

Bulk Annotation

Bulk annotation is the process of annotating all the data with a single click. Users can define the model to be used for this annotation. Once the data is annotated, users can refine the data to ensure the data quality

Annolive supports the use of both publicly available models and your own proprietary models for bulk annotation. If you are using LLMs, annolive lets you to tune the prompt before annotating the whole data

1. Select the Task in the project

In Annolive, a ‘task’ means an annotation task/job. Select the project form projects page. Navigate to the tasks tab in the side bar. Click on the task for annotation

More details on creating the task in Annolive can be found here

2. Configure the models for auto-annotation

Go to the task page

  1. Navigate to the Settings tab within the task view.
  2. In the Settings tab, Activate auto annotation by toggling the enable button.
  3. Choose the required pre-trained model. If you have a custom model, select custom model from the dropdown menu.
  4. If you’ve chosen a custom model, provide the HTTP link to the custom model interface.
Annolive GLiNER integration

3. Bulk Annotate and Refine the Data

Proceed to the Bulk Annotate options in Settings and click the Start button.

Once the bulk annotation process is completed, you can go to individual data items, if necessary, to refine the annotations. Or you can directly export the annotated dataset

References

Thanks for reading. Please contact us for any queries.