eevvgg/gpt2-stance-politic

Model usage

• training procedure available in Colab notebook

# Model usage
from transformers import pipeline

model_path = "eevvgg/gpt2-stance-politic"
cls_task = pipeline(task = "text-classification", model = model_path, tokenizer = model_path)#,  device=0 

sequence = [
            "Looks like a close race with Joe Biden with a slight edge! president  Woodlands, Texas", 
            "I already have them. And that why we need to transition to renwables now! Dependence on liquid fossil fuels make us vulnerable."
]
            
result = cls_task(sequence)

                                        

Metrics

• f1 macro: 77.9

• accuracy: 77.5

Intended uses & limitations

Model suited for classification of stance in social media text. Fine-tuned on a P-Stance dataset of size 3k.