Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 59.91% |
| Test F1 score | 0.6457 |
| Hierarchical loss | 0.91593968 |
| P-adic loss (total) | 253.53761090 |
| P-adic loss (mean) | 0.27618476 |
| Prime base | 79 |
| Number of tags (input features) | 6,936 |
| Non-zero parameters | 2,608 / 2,414,076 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 3,662 |
| Test samples | 918 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 554 | 60.35% | 0.000000 | 0.000000 |
| p^5 | 3 | 0.33% | 0.000000 | 0.000000 |
| p^4 | 11 | 1.20% | 0.000000 | 0.000000 |
| p^3 | 18 | 1.96% | 0.000002 | 0.000037 |
| p^2 | 37 | 4.03% | 0.000160 | 0.005929 |
| p^1 | 42 | 4.58% | 0.012658 | 0.531646 |
| p^0 | 253 | 27.56% | 1.000000 | 253.000000 |
L1 (Lasso) regularization promotes sparsity by driving many coefficients to exactly zero. This model uses ALL available tags (6,936) but L1 regularization selects which features are actually used. The number of non-zero parameters (2,608) indicates how many coefficients the model actually uses.