Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 50.92% |
| Test F1 score | 0.5467 |
| Hierarchical loss | 0.90454930 |
| P-adic loss (total) | 400.33663966 |
| P-adic loss (mean) | 0.26165793 |
| Prime base | 79 |
| Number of tags (input features) | 5,904 |
| Non-zero parameters | 4,328 / 3,253,655 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 6,160 |
| Test samples | 1,530 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 781 | 51.05% | 0.000000 | 0.000000 |
| p^6 | 3 | 0.20% | 0.000000 | 0.000000 |
| p^5 | 4 | 0.26% | 0.000000 | 0.000000 |
| p^4 | 38 | 2.48% | 0.000000 | 0.000001 |
| p^3 | 76 | 4.97% | 0.000002 | 0.000154 |
| p^2 | 125 | 8.17% | 0.000160 | 0.020029 |
| p^1 | 104 | 6.80% | 0.012658 | 1.316456 |
| p^0 | 399 | 26.08% | 1.000000 | 399.000000 |
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 (4,328) indicates how many coefficients the model actually uses.