Aionoscope Foundational Benchmark Dashboard

Loading JSON results…
Loaded JSON files Loading...
Selected models 0

Model Bubble Map

Selected models only, with any supported metric on X, Y, or bubble size

Model Metric Bubble Map
Loading bubble chart settings…

Radar Panels

Absolute median values at the current best layer for each selected model

Categorical Signal AUROC
Per-class median AUROC, excluding constant
Regression By Component Type (R2)
Median across targets per component type
Regression By Parameter Type (R2)
Median across targets per parameter type
Categorical Signal AUPRC
Per-class median AUPRC, excluding constant
Regression By Component Type (Pearson)
Median across targets per component type
Regression By Parameter Type (Pearson)
Median across targets per parameter type

Layer Curves

Median macro metrics across all evaluated layers, with shaded ± std

Macro AUROC By Layer
Categorical probe summary, median line + std band
Macro AUPRC By Layer
Categorical probe summary, median line + std band
Macro Pearson By Layer
Dense probe summary, median line + std band
Macro R2 By Layer
Dense probe summary, median line + std band, clipped at 0 for plotting

Tip: launch a local static server from the repo root with python -m http.server 8000, then open http://localhost:8000/results/dashboard.html.