gliner-datause-catchall-probe

rafmacalaba/gliner-datause-catchall-probe

gliner-datause-catchall-probe is a token classification model from rafmacalaba. Use it when you need labels on individual words, such as names. It is set up for gliner. The card lists the license as apache-2.0.

Multi-origin probe of rafmacalaba/gliner-datause-mentions-catch-all (proposals: NAMEDDATA, DESCRIPTIVEDATA, VAGUEDATA). The Luna v2.4 keep/drop boundary (v2.4-patched-2026-09-05) is read off the frozen representation…

Downloads · 30 days

Access

Public

Updated Sep 8, 2026

Repo size

2.4 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.pt2.4 MB · 62%

At a glance

Task: Token Classification
Library: gliner
License: apache-2.0
Access: Public
Created: Sep 8, 2026
Updated: Sep 8, 2026
SHA: 7e8a0c1d

Try a prompt

Model card

From the Hugging Face model README

gliner-datause-catchall-probe

Multi-origin probe of rafmacalaba/gliner-datause-mentions-catch-all (proposals: NAMED_DATA, DESCRIPTIVE_DATA, VAGUE_DATA). The Luna v2.4 keep/drop boundary (v2.4-patched-2026-09-05) is read off the frozen representation with a small MLP head on [start; end; mean; ±64-token window] span features: extractor proposes, head disposes.

Trained on gold v2 (2,728 Luna-verdict spans over rafmacalaba/datause-extracted-sample, gliner2 config: FCV PADs, JDC operational, refugee PADs, ReliefWeb, JAD PADDY, PRWP) plus 26,618 campaign-v2 Luna single-judge labels drawn origin x specificity x 0.1-decile across the full extraction corpus — 29,346 spans total, doc-disjoint 70/15/15 split, every origin in train, val, and holdout, no passage crosses splits. The full candidate pool (2,729 sample spans + 27,126 campaign spans, passages included) is versioned alongside the splits in rafmacalaba/datause-probe-v3 (candidates config). BCE targets are label-smoothed by 0.1 (eps/2) to curb overconfident scores; scores are ranking-oriented, not calibrated probabilities.

Results (holdout, gold = Luna distillation verdicts)

head AUROC: 0.8896 (distillation fidelity: how well the head carries the Luna v2.4/v2.5 keep/drop boundary)
head best-F1 thr 0.2: precision 0.7550 · recall 0.9350 · f1 0.8354
val AUROC: 0.8893

Per origin

origin	n	keep	AUROC	head best-F1 @ thr
`fcv_pads_east_africa`	1123	431	0.8978	0.7593 @ 0.4
`general_prwp`	1859	1299	0.8635	0.8814 @ 0.3
`jad_paddy_docs`	169	158	0.9522	0.9723 @ 0.2
`jdc_operational`	242	115	0.9086	0.8240 @ 0.2
`refugee_pads`	460	210	0.9248	0.8277 @ 0.3
`reliefweb`	804	447	0.7890	0.7686 @ 0.2

Human check (never trained on)

set	n	keep	AUROC	head best-F1 @ thr
`human473`	473	315	0.8690	0.8629 @ 0.2
`annotator190`	190	110	0.9100	0.8505 @ 0.3
`jdc283`	283	205	0.8240	0.8796 @ 0.1

Labels: gold-v2 verdicts + campaign-v2 Luna single-judge labels (v2.4-patched-2026-09-05, keyed single-span batch, prompt v2.5-draft-2026-09-06); they are a distillation target, not human gold. 7 spans quarantined (unresolved echo) + 1 bad_json excluded — see quarantine.jsonl in rafmacalaba/datause-probe-v3. Head: head.pt; per-span holdout predictions: holdout_predictions.jsonl (key, origin, label, probe_score, head_score); full sweeps: holdout_metrics.json. Human external check: human473_predictions.jsonl (473 annotator190/jdc283 spans, never trained on) with per-set AUROC + best-F1 in holdout_metrics.json → human473 and the table above.

Task: Token Classification
Library: gliner
License: apache-2.0
Created: Sep 8, 2026
Updated: Sep 8, 2026