Silly-Machine/TuPy-Bert-Base-Multilabel

Introduction

TuPy-Bert-Base-Multilabel is a fine-tuned BERT model designed specifically for multilabel classification of hate speech in Portuguese. Derived from the BERTimbau base, TuPy-Bert-Base-Multilabel is a refined solution for addressing categorical hate speech concerns (ageism, aporophobia, body shame, capacitism, LGBTphobia, political, racism, religious intolerance, misogyny, and xenophobia). For more details or specific inquiries, please refer to the BERTimbau repository.

The efficacy of Language Models can exhibit notable variations when confronted with a shift in domain between training and test data. In the creation of a specialized Portuguese Language Model tailored for hate speech classification, the original BERTimbau model underwent fine-tuning processe carried out on the TuPy Hate Speech DataSet, sourced from diverse social networks.

Available models

Model	Arch.	#Layers	#Params
Silly-Machine/TuPy-Bert-Base-Binary-Classifier	BERT-Base	12	109M
Silly-Machine/TuPy-Bert-Large-Binary-Classifier	BERT-Large	24	334M
Silly-Machine/TuPy-Bert-Base-Multilabel	BERT-Base	12	109M
Silly-Machine/TuPy-Bert-Large-Multilabel	BERT-Large	24	334M

Example usage

from transformers import AutoModelForSequenceClassification, AutoTokenizer, AutoConfig
import torch
import numpy as np
from scipy.special import softmax

def classify_hate_speech(model_name, text):
    model = AutoModelForSequenceClassification.from_pretrained(model_name)
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    config = AutoConfig.from_pretrained(model_name)

    # Tokenize input text and prepare model input
    model_input = tokenizer(text, padding=True, return_tensors="pt")

    # Get model output scores
    with torch.no_grad():
        output = model(**model_input)
        scores = softmax(output.logits.numpy(), axis=1)
        ranking = np.argsort(scores[0])[::-1]

    # Print the results
    for i, rank in enumerate(ranking):
        label = config.id2label[rank]
        score = scores[0, rank]
        print(f"{i + 1}) Label: {label} Score: {score:.4f}")

# Example usage
model_name = "Silly-Machine/TuPy-Bert-Base-Multilabel"
text = "Bom dia, flor do dia!!"
classify_hate_speech(model_name, text)