Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 43.15% |
| Test F1 score | 0.4813 |
| Hierarchical loss | 0.88725214 |
| P-adic loss (total) | 1353.58892050 |
| P-adic loss (mean) | 0.29128232 |
| Prime base | 83 |
| Number of tags (input features) | 7,726 |
| Non-zero parameters | 10,511 / 7,688,365 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 18,450 |
| Test samples | 4,647 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 2,005 | 43.15% | 0.000000 | 0.000000 |
| p^6 | 4 | 0.09% | 0.000000 | 0.000000 |
| p^5 | 3 | 0.06% | 0.000000 | 0.000000 |
| p^4 | 90 | 1.94% | 0.000000 | 0.000002 |
| p^3 | 254 | 5.47% | 0.000002 | 0.000444 |
| p^2 | 568 | 12.22% | 0.000145 | 0.082450 |
| p^1 | 374 | 8.05% | 0.012048 | 4.506024 |
| p^0 | 1,349 | 29.03% | 1.000000 | 1349.000000 |
L1 (Lasso) regularization promotes sparsity by driving many coefficients to exactly zero. This model uses ALL available tags (7,726) but L1 regularization selects which features are actually used. The number of non-zero parameters (10,511) indicates how many coefficients the model actually uses.