LaProfeClaudis/LGBeTO_detection_Model

1---2library_name: transformers3base_model: dccuchile/bert-base-spanish-wwm-uncased4tags:5- generated_from_trainer6- hate towards LGBT communities7- BETO8metrics:9- accuracy10- f111- precision12- recall13model-index:14- name: LGBeTO_detection_Model15  results: []16license: cc-by-4.017language:18- es19pipeline_tag: text-classification20---21 22# LGBeTO_detection_Model23 24This is LGBeTO model. Corresponding to a fine-tuned version of [dccuchile/bert-base-spanish-wwm-uncased](https://huggingface.co/dccuchile/bert-base-spanish-wwm-uncased)(Cañete et al., 2023).25It achieves the following results on the evaluation set:26 27- Accuracy: 0.83528- F1: 0.853329- Precision: 0.820530- Recall: 0.888931 32## Authors33 34- **Developed by:** Claudia Martínez-Araneda, Mariella Gutiérrez V., Pedro Gómez M., Diego Maldonado M., Alejandra Segura N., Christian Vidal-Castro35- **Model type:** BERT-based sentiment analysis, BERT-based text classification.36- **Language(s) (NLP):** Spanish37- **License:** CC BY 4.038- **Finetuned from model:** BETO (Cañete et al., 2023)39 40Cite as:41 42@misc{claudia_martínez-araneda_2025,43	author       = { Claudia Martínez-Araneda and Mariella Gutiérrez V. and Pedro Gómez M. and Diego Maldonado M. and Alejandra Segura N. and Christian Vidal-Castro },44	title        = { LGBeTO_detection_Model (Revision a8b5b38) },45	year         = 2025,46	url          = { https://huggingface.co/LaProfeClaudis/LGBeTO_detection_Model },47	doi          = { 10.57967/hf/5406 },48	publisher    = { Hugging Face }49}50 51## Model description52 53LGBeTO was designed to detect discriminatory or hateful language directed toward the LGBTQIA+ community, aiming to support safer and more inclusive online environments.54 55## Intended uses & limitations56 57This model was created for a study conducted strictly for academic and research purposes. The target of hate speech has been anonymised, and there is no intent to harm the perpetrators58in any way. We prioritise protecting the privacy and confidentiality of vulnerable individuals. We carefully remove identifying data, such as user IDs, phone numbers, and addresses, to safeguard privacy before 59sharing the data with our annotators. All data collected comes from public sources. 60 61As authors, we affirm our deep respect for all individuals and explicitly state that we have no intention of prejudicing, biasing, or disrespecting the LGBTQIA+ community or any group. Our work seeks to contribute constructively to inclusive 62and ethical research in artificial intelligence.63 64 65## Training and evaluation data66 67LGBeTO was fine-tuned using comments collected from digital media, such as Twitter, Instagram, websites, and YouTube comments.68The dataset is available in the Zenodo Repository.69 70Cite as:71Martínez-Araneda, C., Maldonado Montiel, D., Gutiérrez Valenzuela, M., Gómez Meneses, P., Segura Navarrete, A.,72& Vidal-Castro, C. (2025). LGBTQIAphobia dataset (augmented and balanced) [Data set]. Zenodo. 73https://doi.org/10.5281/zenodo.1538562274 75## Training procedure76 77- **step 1:** Load the dataSet78- **step 2:** Tokenization and model generation79- **step 3:** Split train-validation80- **step 4:** Training configuration81- **step 5:** Training/Evaluation82 83 84### Training hyperparameters85 86The following hyperparameters were used during training:87- learning_rate: 5e-0588- train_batch_size: 1689- eval_batch_size: 1690- seed: 4291- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments92- num_epochs: 393 94### Training results95 96  | Training Loss | Epoch | Step | Validation Loss | Accuracy | F1     | Precision | Recall |97  |:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:---------:|:------:|98  | 0.4655        | 1.0   | 50   | 0.5517          | 0.755    | 0.7538 | 0.8242    | 0.6944 |99  | 0.1928        | 2.0   | 100  | 0.4830          | 0.825    | 0.8523 | 0.7829    | 0.9352 |100| **0.0718**        | **3.0**    | **150**   | **0.5393**          | **0.835**     | **0.8533**  | **0.8205**    | **0.8889** |101 102 103### Framework versions104 105- Transformers 4.51.3106- Pytorch 2.6.0+cu124107- Datasets 3.6.0108- Tokenizers 0.21.1