Hyperparameter	Value
LANGUAGES	en-es-it
LR	3e-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	True
R_DROP	True
R_KL_ALPHA	1.0
TEXT_NORMALIZE	True

Dev set results (summary)

Metric	Value
f1_macro_dev_0.5	0.6952488687782805
f1_weighted_dev_0.5	0.8306311662921121
accuracy_dev_0.5	0.8129175946547884
f1_macro_dev_best_global	0.7315184893784421
f1_weighted_dev_best_global	0.8708804774663164
accuracy_dev_best_global	0.8730512249443207
f1_macro_dev_best_by_lang	0.7431350114416475
f1_weighted_dev_best_by_lang	0.8703322409830133
accuracy_dev_best_by_lang	0.8663697104677061
default_threshold	0.5
best_threshold_global	0.75
thresholds_by_lang	{"en": 0.75, "it": 0.55, "es": 0.65}

Thresholds

Default: 0.5
Best global: 0.75
Best by language: { "en": 0.75, "it": 0.55, "es": 0.65 }

Detailed evaluation

Classification report @ 0.5

text

              precision    recall  f1-score   support

 no-recl (0)     0.9388    0.8364    0.8846       385
    recl (1)     0.4057    0.6719    0.5059        64

    accuracy                         0.8129       449
   macro avg     0.6722    0.7541    0.6952       449
weighted avg     0.8628    0.8129    0.8306       449

Classification report @ best global threshold (t=0.75)

text

              precision    recall  f1-score   support

 no-recl (0)     0.9205    0.9325    0.9265       385
    recl (1)     0.5593    0.5156    0.5366        64

    accuracy                         0.8731       449
   macro avg     0.7399    0.7240    0.7315       449
weighted avg     0.8690    0.8731    0.8709       449

Classification report @ best per-language thresholds

text

              precision    recall  f1-score   support

 no-recl (0)     0.9333    0.9091    0.9211       385
    recl (1)     0.5270    0.6094    0.5652        64

    accuracy                         0.8664       449
   macro avg     0.7302    0.7592    0.7431       449
weighted avg     0.8754    0.8664    0.8703       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	0.8831	0.5933	0.8787	0.6014	0.5870	0.8746	0.8831
it	163	0.8834	0.8174	0.8855	0.8072	0.8293	0.8882	0.8834
es	132	0.8258	0.7118	0.8381	0.6902	0.7536	0.8587	0.8258

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/october-finetuning-more-variables-sweep-20251012-200701-t06"
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.