Skip to main content

Test Data Generation

Prerequisites

Overview​

Test data generation requires an LLM configuration to generate semantic data correctly.

LLM setup​

You can host the LLM either:

  • Locally in your Kubernetes cluster via the AGENT Helm chart, or
  • On an external server (for example, a managed LLM provider or a self-hosted endpoint).

The following configuration attributes are required:

  • LLM_MODEL_NAME: The model identifier to use for generation.
  • LLM_API_KEY: Credentials used to authenticate with the LLM endpoint (if required).
  • LLM_URL: The base URL of the LLM API endpoint.

Performance considerations​

Generation time depends heavily on the model you choose and the size of the generation request. In practice, model performance is driven by factors such as token throughput, context length, and endpoint concurrency. The LLM model provided with AGENT is Ollama.