Home▸Algorithms & AI▸LLM Evaluation & Benchmarks (2D)

LLM Evaluation & Benchmarks (2D)

2D bar-chart lab: watch benchmark scores and their 95% confidence intervals shift with sample size and noise, and see the composite leaderboard reorder live.

Algorithms & AI2DModerate60 FPS📱 Mobile-adapted⇄ 3D version

This 2D companion reuses the same sampling-noise model as the 3D version — SE = σ/√n, observed = true skill + Gaussian noise — drawn as a flat grouped bar chart with confidence-interval whiskers, so the shrinking-interval effect and the reordering leaderboard are easy to read at a glance.

⚙ Under the hood

Six models, four benchmarks, Box-Muller sampling noise and a live weighted composite leaderboard, rendered on a plain HTML5 canvas.

llm evaluationbenchmarksstatisticsconfidence intervalsampling noisecanvas2d

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install