PCNN Fold 3

Back to PCNN overview · Back to main index

Fold metrics

Metric	Value
Test accuracy	11.45%
Test F1 score	0.0903
Hierarchical loss	0.77610683
P-adic loss (total)	1335.36230000
P-adic loss (mean)	0.72455900
Prime base	79
Hidden layer size	27
Max tags	32
Training samples	7,267
Test samples	1,843

P-adic loss breakdown

Agreement	Count	Share	Cost per mistake	Total contribution
Exact match	211	11.45%	0.000000	0.000000
p^4	1	0.05%	0.000000	0.000000
p^3	64	3.47%	0.000002	0.000130
p^2	127	6.89%	0.000160	0.020349
p^1	106	5.75%	0.012658	1.341772
p^0	1,334	72.38%	1.000000	1334.000000

Tag Rank vs First-Layer Weight Magnitude

Tag rank vs max first-layer weight magnitude — Scatter plot showing the relationship between tag battle ranking and maximum absolute first-layer weight value across all hidden units. Shows which input features contribute most to the parameter constrained neural network's hidden representations.

About p-adic loss

P-adic loss measures the distance between predicted and true taxonomy using p-adic metric (base 79). Lower values indicate closer predictions in the taxonomy hierarchy. This metric is shared with the umllr model for comparison.