Modelpublic

SimoneAstarita/it-no-bio-20251014-t15

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face

0likes5downloads

Model Card

it-no-bio-20251014-t15

Slur reclamation binary classifier Task: LGBTQ+ reclamation vs non-reclamation use of harmful words on social media text.

Trial timestamp (UTC): 2025-10-14 10:47:14 Data case: it

Configuration (trial hyperparameters)

Model: Alibaba-NLP/gte-multilingual-base

Hyperparameter	Value
LANGUAGES	it
LR	2e-05
EPOCHS	3
MAX_LENGTH	256
USE_BIO	False
USELANGTOKEN	False
GATED_BIO	False
FOCAL_LOSS	True
FOCAL_GAMMA	1.5
USE_SAMPLER	True
R_DROP	True
RKLALPHA	0.5
TEXT_NORMALIZE	True

Dev set results (summary)

Metric	Value
f1macrodev_0.5	0.9092483761212495
f1weighteddev_0.5	0.9444368750936943
accuracydev0.5	0.9447852760736196
f1macrodevbestglobal	0.9182957393483709
f1weighteddevbestglobal	0.9502867598443963
accuracydevbest_global	0.950920245398773
f1macrodevbestby_lang	0.9182957393483709
f1weighteddevbestby_lang	0.9502867598443963
accuracydevbestbylang	0.950920245398773
default_threshold	0.5
bestthresholdglobal	0.55
thresholdsbylang	{"it": 0.55}

Thresholds

—Default: 0.5
—Best global: 0.55
—Best by language: { "it": 0.55 }

Detailed evaluation

Classification report @ 0.5

text

              precision    recall  f1-score   support

 no-recl (0)     0.9624    0.9697    0.9660       132
    recl (1)     0.8667    0.8387    0.8525        31

    accuracy                         0.9448       163
   macro avg     0.9145    0.9042    0.9092       163
weighted avg     0.9442    0.9448    0.9444       163

Classification report @ best global threshold (t=0.55)

text

              precision    recall  f1-score   support

 no-recl (0)     0.9627    0.9773    0.9699       132
    recl (1)     0.8966    0.8387    0.8667        31

    accuracy                         0.9509       163
   macro avg     0.9296    0.9080    0.9183       163
weighted avg     0.9501    0.9509    0.9503       163

Classification report @ best per-language thresholds

text

              precision    recall  f1-score   support

 no-recl (0)     0.9627    0.9773    0.9699       132
    recl (1)     0.8966    0.8387    0.8667        31

    accuracy                         0.9509       163
   macro avg     0.9296    0.9080    0.9183       163
weighted avg     0.9501    0.9509    0.9503       163

Per-language metrics (at best-by-lang)

lang	n	acc	f1_macro	f1_weighted	prec_macro	rec_macro	prec_weighted	rec_weighted
it	163	0.9509	0.9183	0.9503	0.9296	0.9080	0.9501	0.9509

Data

—Train/Dev: private multilingual splits with ~15% stratified Dev (by (lang,label)).
—Source: merged EN/IT/ES data with bios retained (ignored if unused by model).

Usage

python

from transformers import AutoTokenizer, AutoModelForSequenceClassification, AutoConfig
import torch, numpy as np

repo = "SimoneAstarita/it-no-bio-20251014-t15"
tok = AutoTokenizer.from_pretrained(repo)
cfg = AutoConfig.from_pretrained(repo)
model = AutoModelForSequenceClassification.from_pretrained(repo)

texts = ["example text ..."]
langs = ["en"]

mode = "best_global"  # or "0.5", "by_lang"

enc = tok(texts, truncation=True, padding=True, max_length=256, return_tensors="pt")
with torch.no_grad():
    logits = model(**enc).logits
probs = torch.softmax(logits, dim=-1)[:, 1].cpu().numpy()

if mode == "0.5":
    th = 0.5
    preds = (probs >= th).astype(int)
elif mode == "best_global":
    th = getattr(cfg, "best_threshold_global", 0.5)
    preds = (probs >= th).astype(int)
elif mode == "by_lang":
    th_by_lang = getattr(cfg, "thresholds_by_lang", {})
    preds = np.zeros_like(probs, dtype=int)
    for lg in np.unique(langs):
        t = th_by_lang.get(lg, getattr(cfg, "best_threshold_global", 0.5))
        preds[np.array(langs) == lg] = (probs[np.array(langs) == lg] >= t).astype(int)
print(list(zip(texts, preds, probs)))

Additional files

reports.json: all metrics (macro/weighted/accuracy) for @0.5, @bestglobal, and @bestbylang. config.json: stores thresholds: defaultthreshold, bestthresholdglobal, thresholdsbylang. postprocessing.json: duplicate threshold info for external tools.