| language: en | |
| license: creativeml-openrail-m | |
| library_name: transformers | |
| tags: | |
| - emotion-classification | |
| - multilabel | |
| - affective-computing | |
| - semantic-affect | |
| - poetic-ai | |
| - BERT | |
| - huggingface | |
| - rosa | |
| - goemotions | |
| datasets: | |
| - go_emotions | |
| metrics: | |
| - f1 | |
| - loss | |
| model_name: Rosa-V1 | |
| model_type: bert | |
| pipeline_tag: text-classification | |
| inference: true | |
| base_model: bert-base-uncased | |
| model_description: | | |
| ROSA (Recursive Ontology of Semantic Affect) is a poetic, fine-tuned BERT-based model | |
| trained on the GoEmotions dataset for multilabel emotion classification. | |
| It classifies text into 28 nuanced emotions plus a neutral category. | |
| Designed by Willinton Triana Cardona, it embodies a graceful fusion of semantic structure, | |
| affective sensitivity, and symbolic embedding. | |
| Eval F1: ~0.5793 | |
| Eval Loss: ~0.0845 | |
| Epochs: 3 | |
| Batch Size: 16 (train), 32 (eval) | |
| 🌹 This is the Rosa of Barcelona— my first blossom of poetic logic and emotional resonance in AI. | |
| author: | |
| name: Willinton Triana Cardona | |
| role: creator | |
| description: Philosopher · AI Researcher · Poetic Architect of Emotion-Aware Systems | |