Sign in

Relevance testing runs across a user-specified sample size configurable from 1 to 100 samples per prompt, defaulting to 1 for manual runs and 10 samples during automated workflows to capture response variance.

  1. 1 Probe execution and threading architecture (recommended)
  2. 2 Prompt design and binary response parsing
  3. 3 Storage in probe_results and DB aggregation

Referenced by