A Python tool for analyzing and visualizing differences between two Aider benchmark runs through an interactive dashboard.
- Compares two benchmark run folders (raw or tar-bz2 archives)
- Interactive web dashboard built with Dash
- Detailed test-by-test comparison analysis
- Visualizes performance changes and differences
- Test result comparison
- Performance metrics visualization
- Interactive data filtering and sorting
- Detailed test case analysis
pipx install cedarverse-bda# Info on a single benchmark run
bda --info <run-path>
# Delta Analysis
bda <run1-path> [run2-path]
# Interactive dashboard
bda --dashboard <run1-path> [run2-path]
# Install development dependencies
pip install -e .[dev]