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

Dev set results (summary)

Metric	Value
f1_macro_dev_0.5	0.7308074106835716
f1_weighted_dev_0.5	0.866252272018095
accuracy_dev_0.5	0.8641425389755011
f1_macro_dev_best_global	0.7588614393125672
f1_weighted_dev_best_global	0.8874859755178593
accuracy_dev_best_global	0.8930957683741648
f1_macro_dev_best_by_lang	0.7424440224701321
f1_weighted_dev_best_by_lang	0.872443990751573
accuracy_dev_best_by_lang	0.8708240534521158
default_threshold	0.5
best_threshold_global	0.55
thresholds_by_lang	{"en": 0.4, "it": 0.55, "es": 0.5}

Thresholds

Default: 0.5
Best global: 0.55
Best by language: { "en": 0.4, "it": 0.55, "es": 0.5 }

Detailed evaluation

Classification report @ 0.5

text

              precision    recall  f1-score   support

 no-recl (0)     0.9263    0.9143    0.9203       385
    recl (1)     0.5217    0.5625    0.5414        64

    accuracy                         0.8641       449
   macro avg     0.7240    0.7384    0.7308       449
weighted avg     0.8686    0.8641    0.8663       449

Classification report @ best global threshold (t=0.55)

text

              precision    recall  f1-score   support

 no-recl (0)     0.9223    0.9558    0.9388       385
    recl (1)     0.6600    0.5156    0.5789        64

    accuracy                         0.8931       449
   macro avg     0.7912    0.7357    0.7589       449
weighted avg     0.8849    0.8931    0.8875       449

Classification report @ best per-language thresholds

text

              precision    recall  f1-score   support

 no-recl (0)     0.9291    0.9195    0.9243       385
    recl (1)     0.5441    0.5781    0.5606        64

    accuracy                         0.8708       449
   macro avg     0.7366    0.7488    0.7424       449
weighted avg     0.8743    0.8708    0.8724       449

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

lang	n	acc	f1_macro	f1_weighted	prec_macro	rec_macro	prec_weighted	rec_weighted
en	154	0.8506	0.5331	0.8531	0.5321	0.5344	0.8557	0.8506
it	163	0.9202	0.8618	0.9175	0.8902	0.8397	0.9175	0.9202
es	132	0.8333	0.7199	0.8441	0.6985	0.7580	0.8617	0.8333

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/trilingual-no-bio-20251014-t03"
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, @best_global, and @best_by_lang. config.json: stores thresholds: default_threshold, best_threshold_global, thresholds_by_lang. postprocessing.json: duplicate threshold info for external tools.