Macro = mean of 13 category scores (each weighs 1/13). Micro = mean over all 100 tasks. Every score cell is heat-shaded on one red→yellow→green scale, like the benchmark heatmap.