es-no-bio-sweep-20251013-174220-t19
Slur reclamation binary classifier
Task: LGBTQ+ reclamation vs non-reclamation use of harmful words on social media text.
Trial timestamp (UTC): 2025-10-13 17:42:20
Data case:
es
Configuration (trial hyperparameters)
Model: Alibaba-NLP/gte-multilingual-base
| Hyperparameter | Value |
|---|---|
| LANGUAGES | es |
| LR | 3e-05 |
| EPOCHS | 3 |
| MAX_LENGTH | 256 |
| USE_BIO | False |
| USE_LANG_TOKEN | False |
| GATED_BIO | False |
| FOCAL_LOSS | True |
| FOCAL_GAMMA | 2.5 |
| 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 | 0.6549019607843137 |
| f1_weighted_dev_0.5 | 0.786096256684492 |
| accuracy_dev_0.5 | 0.7575757575757576 |
| f1_macro_dev_best_global | 0.7105263157894737 |
| f1_weighted_dev_best_global | 0.8572567783094099 |
| accuracy_dev_best_global | 0.8636363636363636 |
| f1_macro_dev_best_by_lang | 0.7105263157894737 |
| f1_weighted_dev_best_by_lang | 0.8572567783094099 |
| accuracy_dev_best_by_lang | 0.8636363636363636 |
| default_threshold | 0.5 |
| best_threshold_global | 0.6 |
| thresholds_by_lang | {"es": 0.6} |
Thresholds
- Default:
0.5 - Best global:
0.6 - Best by language:
{ "es": 0.6 }
Detailed evaluation
Classification report @ 0.5
precision recall f1-score support
no-recl (0) 0.9348 0.7679 0.8431 112
recl (1) 0.3500 0.7000 0.4667 20
accuracy 0.7576 132
macro avg 0.6424 0.7339 0.6549 132
weighted avg 0.8462 0.7576 0.7861 132
Classification report @ best global threshold (t=0.60)
precision recall f1-score support
no-recl (0) 0.9052 0.9375 0.9211 112
recl (1) 0.5625 0.4500 0.5000 20
accuracy 0.8636 132
macro avg 0.7338 0.6937 0.7105 132
weighted avg 0.8533 0.8636 0.8573 132
Classification report @ best per-language thresholds
precision recall f1-score support
no-recl (0) 0.9052 0.9375 0.9211 112
recl (1) 0.5625 0.4500 0.5000 20
accuracy 0.8636 132
macro avg 0.7338 0.6937 0.7105 132
weighted avg 0.8533 0.8636 0.8573 132
Per-language metrics (at best-by-lang)
| lang | n | acc | f1_macro | f1_weighted | prec_macro | rec_macro | prec_weighted | rec_weighted |
|---|---|---|---|---|---|---|---|---|
| es | 132 | 0.8636 | 0.7105 | 0.8573 | 0.7338 | 0.6937 | 0.8533 | 0.8636 |
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
from transformers import AutoTokenizer, AutoModelForSequenceClassification, AutoConfig
import torch, numpy as np
repo = "SimoneAstarita/es-no-bio-sweep-20251013-174220-t19"
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.