Back to ULR overview · Back to main index
| Metric | Value |
|---|---|
| Test accuracy | 42.85% |
| Test F1 score | 0.4723 |
| Hierarchical loss | 0.88947566 |
| P-adic loss (total) | 1316.47484295 |
| P-adic loss (mean) | 0.28421305 |
| Prime base | 83 |
| Number of tags (input features) | 7,726 |
| Non-zero parameters | 10,651 / 7,672,911 (99.9% sparse) |
| L1 regularization (C) | 1.0000 |
| Training samples | 18,465 |
| Test samples | 4,632 |
| Agreement | Count | Share | Cost per mistake | Total contribution |
|---|---|---|---|---|
| Exact match | 1,985 | 42.85% | 0.000000 | 0.000000 |
| p^6 | 2 | 0.04% | 0.000000 | 0.000000 |
| p^5 | 8 | 0.17% | 0.000000 | 0.000000 |
| p^4 | 85 | 1.84% | 0.000000 | 0.000002 |
| p^3 | 264 | 5.70% | 0.000002 | 0.000462 |
| p^2 | 612 | 13.21% | 0.000145 | 0.088837 |
| p^1 | 364 | 7.86% | 0.012048 | 4.385542 |
| p^0 | 1,312 | 28.32% | 1.000000 | 1312.000000 |
L1 (Lasso) regularization promotes sparsity by driving many coefficients to exactly zero. This model uses ALL available tags (7,726) but L1 regularization selects which features are actually used. The number of non-zero parameters (10,651) indicates how many coefficients the model actually uses.