| system | n | chrf | edit_sim | rougeL | exact_match | number_recall | placeholders_exact |
|---|---|---|---|---|---|---|---|
| B0_identity | 2493 | 69.760 [68.643, 70.936] | 0.689 [0.678, 0.700] | 0.697 [0.684, 0.711] | 0.000 [0.000, 0.000] | 1.000 [1.000, 1.000] | 1.000 [1.000, 1.000] |
| B1_generic_paraphrase | 2493 | 67.007 [65.984, 68.139] | 0.576 [0.565, 0.588] | 0.614 [0.600, 0.628] | 0.000 [0.000, 0.000] | 0.999 [0.998, 1.000] | 1.000 [1.000, 1.000] |
| htr-m1-small | 2493 | 69.410 [68.354, 70.503] | 0.681 [0.671, 0.692] | 0.690 [0.677, 0.703] | 0.000 [0.000, 0.000] | 0.985 [0.979, 0.991] | 0.989 [0.982, 0.995] |
| htr-m1-tiny-s2 | 2493 | 69.741 [68.705, 70.834] | 0.689 [0.679, 0.699] | 0.697 [0.684, 0.710] | 0.000 [0.000, 0.000] | 0.986 [0.979, 0.991] | 0.991 [0.986, 0.995] |
| htr-m1-tiny-s3 | 2493 | 69.664 [68.623, 70.755] | 0.686 [0.676, 0.697] | 0.694 [0.681, 0.707] | 0.000 [0.000, 0.000] | 0.985 [0.977, 0.991] | 0.994 [0.988, 0.998] |
| htr-m1-tiny | 2493 | 69.652 [68.607, 70.760] | 0.685 [0.676, 0.695] | 0.693 [0.681, 0.706] | 0.000 [0.000, 0.000] | 0.985 [0.978, 0.991] | 0.988 [0.978, 0.995] |
| htr-m2-tiny | 2493 | 69.458 [68.382, 70.572] | 0.684 [0.674, 0.695] | 0.693 [0.680, 0.706] | 0.000 [0.000, 0.000] | 0.985 [0.978, 0.990] | 0.986 [0.978, 0.992] |
| htr-m3light-tiny | 2493 | 69.257 [68.172, 70.377] | 0.683 [0.673, 0.693] | 0.691 [0.678, 0.704] | 0.000 [0.000, 0.000] | 0.982 [0.974, 0.988] | 0.987 [0.978, 0.994] |
| htr-m3tokenonly-tiny-s2 | 2493 | 69.497 [68.449, 70.573] | 0.685 [0.676, 0.696] | 0.694 [0.682, 0.707] | 0.000 [0.000, 0.000] | 0.986 [0.979, 0.992] | 0.990 [0.983, 0.996] |
| htr-m3tokenonly-tiny | 2493 | 69.650 [68.584, 70.759] | 0.687 [0.677, 0.698] | 0.695 [0.682, 0.709] | 0.000 [0.000, 0.000] | 0.985 [0.978, 0.990] | 0.986 [0.979, 0.993] |
Stylometric W1 (mean over features, standardised; lower = closer to originals):
- rewrite vs original: 0.160
- B0_identity vs original: 0.160 · probe P(original) mean 0.42
- B1_generic_paraphrase vs original: 0.214 · probe P(original) mean 0.48
- htr-m1-small vs original: 0.097 · probe P(original) mean 0.51
- htr-m1-tiny-s2 vs original: 0.100 · probe P(original) mean 0.50
- htr-m1-tiny-s3 vs original: 0.101 · probe P(original) mean 0.51
- htr-m1-tiny vs original: 0.103 · probe P(original) mean 0.50
- htr-m2-tiny vs original: 0.101 · probe P(original) mean 0.51
- htr-m3light-tiny vs original: 0.105 · probe P(original) mean 0.51
- htr-m3tokenonly-tiny-s2 vs original: 0.103 · probe P(original) mean 0.50
- htr-m3tokenonly-tiny vs original: 0.107 · probe P(original) mean 0.50