PCNN Fold 4

Back to PCNN overview · Back to main index

Fold metrics

Metric	Value
Test accuracy	13.49%
Test F1 score	0.0969
Hierarchical loss	0.77188940
P-adic loss (total)	1109.86230000
P-adic loss (mean)	0.73745000
Prime base	79
Hidden layer size	27
Max tags	32
Training samples	6,185
Test samples	1,505

P-adic loss breakdown

Agreement	Count	Share	Cost per mistake	Total contribution
Exact match	203	13.49%	0.000000	0.000000
p^4	2	0.13%	0.000000	0.000000
p^3	36	2.39%	0.000002	0.000073
p^2	88	5.85%	0.000160	0.014100
p^1	67	4.45%	0.012658	0.848101
p^0	1,109	73.69%	1.000000	1109.000000

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.