alissonf216/parliamentary-bertimbau-auditor

Parliamentary BERTimbau Auditor

Fine-tuned neuralmind/bert-base-portuguese-cased for 4-class offensive / hate-speech detection in Brazilian parliamentary discourse.

Labels

id	label
0	NEUTRAL
1	GENERIC_OFFENSE
2	TARGETED_OFFENSE
3	EXPLICIT_HATE_SPEECH

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

repo = "parliamentary-bertimbau-auditor"  # or your HF repo id
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForSequenceClassification.from_pretrained(repo)
text = "Senhor presidente, peço a palavra."
inputs = tok(text, return_tensors="pt", truncation=True, max_length=128)
with torch.no_grad():
    probs = torch.softmax(model(**inputs).logits, dim=-1)[0]
pred = int(probs.argmax())
print(model.config.id2label[pred], float(probs[pred]))

Best validation Macro-F1 (this run): 0.3970