Latest Benchmark

Rolling nightly benchmark compiled from the live operational runs. Snapshot label: latest-2026-10-12T0017Z.

Trained params is the average non-zero parameter count in the fitted model across folds. Avg active params / classification is the mean number of active parameters or scoring decisions touched while classifying one product.

20,227Filtered products

6,134Filtered tags

805Filtered taxonomies

83Prime base

View UMLLR tag-order ablation View operational UMLLR page

Benchmark dashboard chart for latest benchmark view — Small-multiples dashboard generated directly from the comparison table rows below.

Model comparison chart for latest benchmark view — Log-log scatter of trained parameters versus mean p-adic loss, overlaid with the current parsimoniousness baseline.

Active parameter comparison chart for latest benchmark view — Scatter plot of avg active params versus mean p-adic loss, excluding PCLR and PCNN. Fitted equation: mean p-adic loss = -0.0826 log10(active params) + 0.4110. Regression fitted on log10(active params): R²=0.670, p=0.024.

Active parameter versus log-loss chart for latest benchmark view — Scatter plot of avg active params versus log10(mean p-adic loss), excluding PCLR and PCNN. Fitted equation: log10(mean p-adic loss) = -0.1313 log10(active params) - 0.4076. Regression fitted on log10(active params): R²=0.806, p=0.006.

Model	Trained params	Mean p-adic loss	Exact acc.	Prefix-2 acc.	Avg active params / classification
Dummy Baseline	1.0	0.595174	3.92%	20.50%	1.00
Importance-Optimised p-adic Linear Regression	1816.0	0.309062	45.21%	60.01%	1.12
Parameter-constrained Logistic Regression	5615.4	0.898426	1.07%	5.70%	905.09
Unconstrained Logistic Regression with L1	11955.4	0.134469	60.97%	77.62%	825.99
Parameter-constrained Neural Network	5188.2	0.534066	13.31%	31.84%	4821.82
Decision Tree	107725.4	0.174072	53.57%	73.40%	626.17
Level-wise Logistic Regression	387799.2	0.154893	45.85%	74.54%	249.59
Unconstrained Neural Network with L2	84048.6	0.146826	58.66%	76.85%	10507.43
Zubarev (greedy init.)	1993.4	0.344849	41.13%	56.48%	1.38