Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 51.80% |
| Test F1 score | 0.5612 |
| Hierarchical loss | 0.95618480 |
| P-adic loss (total) | 254.42508136 |
| P-adic loss (mean) | 0.20854515 |
| Prime base | 71 |
| Number of tags (input features) | 2,797 |
| Non-zero parameters | 3,070 / 1,029,664 (99.7% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 4,961 |
| Test samples | 1,220 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 632 | 51.80% | 0.000000 | 0.000000 |
| p^6 | 2 | 0.16% | 0.000000 | 0.000000 |
| p^5 | 0 | 0.00% | 0.000000 | 0.000000 |
| p^4 | 21 | 1.72% | 0.000000 | 0.000001 |
| p^3 | 59 | 4.84% | 0.000003 | 0.000165 |
| p^2 | 154 | 12.62% | 0.000198 | 0.030549 |
| p^1 | 99 | 8.11% | 0.014085 | 1.394366 |
| p^0 | 253 | 20.74% | 1.000000 | 253.000000 |
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 (3,070) indicates how many coefficients the model actually uses.