llm-benchmark - LLM Benchmark Design and Optimization
Guide for designing, running, and interpreting LLM benchmark experiments with statistical analysis and data leakage prevention
Tags
Updated: 2026-05-28Capabilities
Typical Inputs
Typical Outputs
What this skill does
- design test suites
- run benchmark experiments
- compare prompt variants
- analyze pass rates
- calculate confidence intervals
- detect data leakage
- measure per-turn metrics
- interpret test results
- optimize prompts
- validate test constraints
Inputs
- test dataset
- prompt configuration
- LLM model
- test suite
- statistical parameters
- target effect size
Outputs
- test results
- pass rate
- per-turn metrics
- statistical analysis
- performance comparison
Requirements
- LLM model access
- statistical analysis tools
- test environment
- test data access
