Skip to main content
Version: v0.3.0


H2O Label Genie is an application that enables you with zero-shot learning models to rapidly label your datasets for annotation tasks in computer vision (CV), natural language processing (NLP), and audio; labeled datasets are supported in H2O Hydrogen Torch.

The below sections provide answers to frequently asked questions. If you have additional questions, please send them to


How can I upload my data (dataset) to H2O Label Genie?

Before you can import your dataset to H2O Label Genie, your dataset must meet the following two requirements:

  1. The dataset data type needs to be text,image, or audio.
  2. The dataset (data) needs to be in a zip (for image and audio data) or CSV file (for text data).

After meeting the above two requirements, see the following section to learn how to import your dataset: Import a dataset.

What is the difference between a CPU and GPU bundle of H2O Label Genie?

With H2O Label Genie v0.3.0+, you can obtain a CPU or GPU-based bundle of the application. H2O Label Genie offers a GPU-based bundle to users with sufficient resources to experience better performance of supported zero-shot models and clustering tasks. In the case of clustering tasks, a GPU-based bundle accelerates the creation of the 2D and 3D embeddings.