Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 50.69% |
| Test F1 score | 0.5509 |
| Hierarchical loss | 0.95517216 |
| P-adic loss (total) | 246.73417776 |
| P-adic loss (mean) | 0.20010882 |
| Prime base | 71 |
| Number of tags (input features) | 2,797 |
| Non-zero parameters | 3,231 / 1,021,270 (99.7% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 4,948 |
| Test samples | 1,233 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 625 | 50.69% | 0.000000 | 0.000000 |
| p^6 | 1 | 0.08% | 0.000000 | 0.000000 |
| p^5 | 0 | 0.00% | 0.000000 | 0.000000 |
| p^4 | 21 | 1.70% | 0.000000 | 0.000001 |
| p^3 | 70 | 5.68% | 0.000003 | 0.000196 |
| p^2 | 150 | 12.17% | 0.000198 | 0.029756 |
| p^1 | 121 | 9.81% | 0.014085 | 1.704225 |
| p^0 | 245 | 19.87% | 1.000000 | 245.000000 |
L1 (Lasso) regularization promotes sparsity by driving many coefficients to exactly zero. This model uses ALL available tags (2,797) but L1 regularization selects which features are actually used. The number of non-zero parameters (3,231) indicates how many coefficients the model actually uses.