Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 53.43% |
| Test F1 score | 0.5624 |
| Hierarchical loss | 0.90723142 |
| P-adic loss (total) | 397.09641926 |
| P-adic loss (mean) | 0.25937062 |
| Prime base | 79 |
| Number of tags (input features) | 5,904 |
| Non-zero parameters | 4,447 / 3,253,655 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 6,159 |
| Test samples | 1,531 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 818 | 53.43% | 0.000000 | 0.000000 |
| p^6 | 3 | 0.20% | 0.000000 | 0.000000 |
| p^5 | 7 | 0.46% | 0.000000 | 0.000000 |
| p^4 | 36 | 2.35% | 0.000000 | 0.000001 |
| p^3 | 59 | 3.85% | 0.000002 | 0.000120 |
| p^2 | 127 | 8.30% | 0.000160 | 0.020349 |
| p^1 | 85 | 5.55% | 0.012658 | 1.075949 |
| p^0 | 396 | 25.87% | 1.000000 | 396.000000 |
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 (4,447) indicates how many coefficients the model actually uses.