Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 53.02% |
| Test F1 score | 0.5556 |
| Hierarchical loss | 0.90616631 |
| P-adic loss (total) | 407.18232930 |
| P-adic loss (mean) | 0.26457591 |
| Prime base | 79 |
| Number of tags (input features) | 5,904 |
| Non-zero parameters | 4,388 / 3,259,560 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 6,151 |
| Test samples | 1,539 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 820 | 53.28% | 0.000000 | 0.000000 |
| p^6 | 2 | 0.13% | 0.000000 | 0.000000 |
| p^5 | 1 | 0.06% | 0.000000 | 0.000000 |
| p^4 | 36 | 2.34% | 0.000000 | 0.000001 |
| p^3 | 72 | 4.68% | 0.000002 | 0.000146 |
| p^2 | 110 | 7.15% | 0.000160 | 0.017625 |
| p^1 | 92 | 5.98% | 0.012658 | 1.164557 |
| p^0 | 406 | 26.38% | 1.000000 | 406.000000 |
L1 (Lasso) regularization promotes sparsity by driving many coefficients to exactly zero. This model uses ALL available tags (5,904) but L1 regularization selects which features are actually used. The number of non-zero parameters (4,388) indicates how many coefficients the model actually uses.