taxonomy_association
Taxonomy-peaked tags first
For each tag, measure the share of its training occurrences that land in its single most common taxonomy. Tags with the strongest one-taxonomy association are scored first.
← Back to Latest benchmark summary
Rolling nightly benchmark compiled from the live operational runs. Best strategy in this bundle: taxonomy_association (0.237259 mean p-adic loss).
For these UMLLR ablations, Avg active params / classification is the mean number of active coefficients touched while classifying one product.
Ordering methods
The ablation keeps the greedy p-adic regressor fixed and changes only the tag ordering heuristic used before coefficient fitting.
frequency
Most common tags first
Ranks tags by how often they appear in the training products.
battle_elo
Pairwise battle ranking
Ranks tags by fold-local Elo scores estimated from tag battles, while excluding the holdout fold from the ranking fit.
mean_title_position
Average title position
Ranks tags by their average recorded title position in the training products.
random
Seeded random control
Uses a seeded random shuffle of the training tag vocabulary as a control condition.
| Strategy | Mean p-adic loss | Δ vs battle_elo | Fold wins | Exact acc. | Prefix-2 acc. | Avg active params / classification |
|---|---|---|---|---|---|---|
| taxonomy_association | 0.237259 | -0.023754 | 5/5 | 68.15% | 60.65% | 0.84 |
| frequency | 0.250888 | -0.010124 | 5/5 | 63.51% | 59.41% | 1.20 |
| battle_elo | 0.261013 | 0.000000 | 0/5 | 64.88% | 58.17% | 0.98 |
| mean_title_position | 0.269118 | 0.008106 | 0/5 | 62.23% | 57.06% | 1.16 |
| random | 0.294237 | 0.033225 | 1/25 | 61.02% | 55.13% | 1.04 |