Skip to main content

Metrics

The Metrics page displays evaluation metrics for uploaded datasets (queries and responses, with or without context) and user interactions, enabling the AI team to assess the quality of the LLM app in both pre-production and production environments. Key metrics include:

  • Helpfulness: Overall utility of the response relative to the prompt.
  • Correctness: Accuracy and inclusion of relevant facts.
  • Coherence: Clarity and consistency of expression.
  • Complexity: Depth of expertise required for the response.
  • Verbosity: Appropriateness of detail based on the prompt.

Additional metrics (Context Adherence, Truthfulness, Completeness, Safety) coming soon.

This page allows engineers and product managers to monitor and enhance response quality across various stages.