francesco-zatto/twitter-roberta-base-hate-freeze-embeddings-weighted-L-sexism-detector

RoBERTa Sexism Classifier (Freeze Embeddings / Weighted Loss)

This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-hate, trained for multi-class sexism detection on the EXIST 2023 Task 2 dataset.

Experiment Details: freeze_embeddings

This repository contains the Freeze Embeddings variant of our ablation study.

• —Frozen Embeddings: Only the base embedding layer (model.roberta.embeddings.parameters()) was frozen during training. All subsequent transformer encoder layers and the classification head were fully fine-tuned. This allows the model to learn task-specific attention patterns while anchoring the fundamental word representations.
• —Weighted Loss: Because the EXIST 2023 dataset contains class imbalances, training was conducted using a weighted Cross-Entropy loss function. This ensures the model does not become heavily biased toward the majority class (e.g., Non-sexist) and adequately penalizes errors on the minority classes.

Intended Use

Categorizes English tweets into one of four sexist intentions:

1. 1.- (Non-sexist)
2. 2.DIRECT (Directly sexist messages)
3. 3.JUDGEMENTAL (Messages condemning sexist behaviors)
4. 4.REPORTED (Messages reporting a sexist situation)

Preprocessing

Inputs must be preprocessed to match the CardiffNLP base model formatting. Note that this model is case-sensitive, so do not aggressively lowercase your text if you want to preserve capitalization signals:

• —Replace user mentions (@user) with the token @user
• —Replace URLs with the token http

Evaluation Results (Test Set)

• —Macro F1: 0.5508
• —Precision: 0.5315
• —Recall: 0.5924

How to Use

from transformers import AutoTokenizer, AutoModelForSequenceClassification

repo_id = "francesco-zatto/twitter-roberta-base-hate-freeze-embeddings-weighted-L-sexism-detector"
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoModelForSequenceClassification.from_pretrained(repo_id)

inputs = tokenizer("Your cleaned tweet text here", return_tensors="pt")
outputs = model(**inputs)