Model Trained

• Problem type: Classificação de sentimentos em dataset interno do Sebrae RS
• Model ID: 96390146647
• CO2 Emissions (in grams): 0.6308
• "id2label": {"0": "Negativo", "1": "Neutro", "2": "Positivo"}

Validation Metrics

• Loss: 0.143
• Accuracy: 0.965
• Macro F1: 0.935
• Micro F1: 0.965
• Weighted F1: 0.964
• Macro Precision: 0.938
• Micro Precision: 0.965
• Weighted Precision: 0.964
• Macro Recall: 0.933
• Micro Recall: 0.965
• Weighted Recall: 0.965

Usage

Python API:

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("ggrazzioli/cls_sentimento_sebrae")

tokenizer = AutoTokenizer.from_pretrained("ggrazzioli/cls_sentimento_sebrae")

inputs = tokenizer("Gostei muito dos serviços gerados, recomendo a todos!", return_tensors="pt")

outputs = model(**inputs)