SecurePII-Bench
Collection
262 items • Updated
study string | base_model string | train_flags string | eval_flags string | datasets dict | total_runs int64 | max_train_samples null | max_eval_samples null | runs dict | summary_table list | latex_row string | generated_at timestamp[s] | note string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
set-consistent across NER datasets | fastino/gliner2-multi-v1 | --loss set | --threshold 0.5 --use-desc --schema-case original --no-extra-val | {
"fin": {
"seeds": [
42,
43,
44,
45,
46
],
"epochs": 8,
"batch": 16,
"patience": 2,
"max_val": null,
"repo": "quynong/fin",
"labels": [
"PER",
"ORG",
"LOC",
"MISC"
],
"eval_split": "test"
},
"mit_restaurant": {
"see... | 18 | null | null | {
"fin-set-seed42": {
"status": "ok",
"dataset": "fin",
"config": "set",
"seed": 42,
"repo": "AITeamUIT/gliner2-ner-fin-set-seed42",
"train_flags": "--loss set",
"eval_flags": "--threshold 0.5 --use-desc --schema-case original --no-extra-val",
"epochs": 8,
"batch_size": 16,
"pa... | [
{
"mode": "lenient",
"dataset": "fin",
"n_seeds": 5,
"micro_f1_mean": 80.93,
"micro_f1_std": 2.4,
"micro_f1_sem": 1.07,
"macro_f1_mean": 45.83,
"macro_f1_std": 8.08,
"macro_f1_sem": 3.61,
"micro_p_mean": 85.29,
"micro_r_mean": 77.4,
"micro_f1": "80.93 +/- 2.40",
"... | \textbf{Set-consistent (ours)} & $ 86.91_{\pm 0.74} $ & $ 76.96_{\pm 0.40} $ & $ 40.56_{\pm 5.90} $ & $ 81.46_{\pm 0.44} $ & \textbf{71.47} \\ | 2026-08-22T09:38:00 | Dong Set-consistent cho tab:macro-f1 (strict macro F1). Thay dong cu 88.00/78.28/70.98/81.63 — do la 1 seed, eval non-desc, giao thuc cu. Run nay eval --use-desc --schema-case original, threshold 0.5 (null anchor). |
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Set-consistent — 4 bo NER, 18 run
- base model:
fastino/gliner2-multi-v1 - train:
--loss set(mac dinh cua train.py cho moi tham so set) - eval:
--threshold 0.5 --use-desc --schema-case original --no-extra-val
--loss settu bat learnable null anchor trong scorer, nen checkpoint decode o threshold 0.5 — khac 0.7 cua bang loss-ablation 5 config. Khac biet co chu dich, khong phai threshold tuning.
Ngan sach tung bo
| bo | repo | nhan | split | seeds | epochs | batch |
|---|---|---|---|---|---|---|
fin |
quynong/fin |
4 | test | 5 | 8 | 16 |
mit_restaurant |
quynong/mit_restaurant |
8 | test | 5 | 6 | 16 |
conll2003 |
quynong/conll2003 |
4 | test | 5 | 6 | 16 |
ontonotes5 |
quynong/ontonotes5 |
18 | test | 3 | 3 | 16 |
Ket qua (%, mean +/- std tren seed)
| mode | bo | n seeds | macro F1 | micro F1 | micro P | micro R |
|---|---|---|---|---|---|---|
| lenient | fin | 5 | 45.83 +/- 8.08 | 80.93 +/- 2.40 | 85.29 | 77.4 |
| lenient | mit_restaurant | 5 | 88.14 +/- 0.38 | 89.09 +/- 0.25 | 90.19 | 88.02 |
| lenient | conll2003 | 5 | 88.99 +/- 0.78 | 90.29 +/- 0.80 | 90.69 | 89.89 |
| lenient | ontonotes5 | 3 | 81.13 +/- 0.24 | 90.53 +/- 0.10 | 89.42 | 91.68 |
| strict | fin | 5 | 40.56 +/- 5.90 | 75.41 +/- 1.86 | 79.55 | 72.06 |
| strict | mit_restaurant | 5 | 81.46 +/- 0.44 | 82.45 +/- 0.35 | 83.47 | 81.45 |
| strict | conll2003 | 5 | 86.91 +/- 0.74 | 88.64 +/- 0.80 | 89.05 | 88.25 |
| strict | ontonotes5 | 3 | 76.96 +/- 0.40 | 87.42 +/- 0.15 | 86.34 | 88.52 |
Dong cho tab:macro-f1 (strict macro F1)
\textbf{Set-consistent (ours)} & $ 86.91_{\pm 0.74} $ & $ 76.96_{\pm 0.40} $ & $ 40.56_{\pm 5.90} $ & $ 81.46_{\pm 0.44} $ & \textbf{71.47} \\
Checkpoint tung run
fin-set-seed42-> AITeamUIT/gliner2-ner-fin-set-seed42 (strict macro F1 47.61)fin-set-seed43-> AITeamUIT/gliner2-ner-fin-set-seed43 (strict macro F1 45.63)fin-set-seed44-> AITeamUIT/gliner2-ner-fin-set-seed44 (strict macro F1 39.22)fin-set-seed45-> AITeamUIT/gliner2-ner-fin-set-seed45 (strict macro F1 36.50)fin-set-seed46-> AITeamUIT/gliner2-ner-fin-set-seed46 (strict macro F1 33.82)mit_restaurant-set-seed42-> AITeamUIT/gliner2-ner-mit_restaurant-set-seed42 (strict macro F1 81.44)mit_restaurant-set-seed43-> AITeamUIT/gliner2-ner-mit_restaurant-set-seed43 (strict macro F1 81.92)mit_restaurant-set-seed44-> AITeamUIT/gliner2-ner-mit_restaurant-set-seed44 (strict macro F1 80.74)mit_restaurant-set-seed45-> AITeamUIT/gliner2-ner-mit_restaurant-set-seed45 (strict macro F1 81.52)mit_restaurant-set-seed46-> AITeamUIT/gliner2-ner-mit_restaurant-set-seed46 (strict macro F1 81.70)conll2003-set-seed42-> AITeamUIT/gliner2-ner-conll2003-set-seed42 (strict macro F1 87.29)conll2003-set-seed43-> AITeamUIT/gliner2-ner-conll2003-set-seed43 (strict macro F1 85.94)conll2003-set-seed44-> AITeamUIT/gliner2-ner-conll2003-set-seed44 (strict macro F1 86.78)conll2003-set-seed45-> AITeamUIT/gliner2-ner-conll2003-set-seed45 (strict macro F1 86.62)conll2003-set-seed46-> AITeamUIT/gliner2-ner-conll2003-set-seed46 (strict macro F1 87.90)ontonotes5-set-seed42-> AITeamUIT/gliner2-ner-ontonotes5-set-seed42 (strict macro F1 77.06)ontonotes5-set-seed43-> AITeamUIT/gliner2-ner-ontonotes5-set-seed43 (strict macro F1 77.31)ontonotes5-set-seed44-> AITeamUIT/gliner2-ner-ontonotes5-set-seed44 (strict macro F1 76.52)
generated 2026-08-22T09:38:00
- Downloads last month
- 15