Stance Classification

---

Microsoft DeBERTa (Fine-Tuned)

Training Summary:

• global_step: 375
• epoch: 3.0
• training_loss: 0.93199
  • loss_type: 0.011921
  • loss_stance: 0.832819

Evaluation Metrics:

• test_loss: 0.8482
• test_runtime: 94.79 seconds

Classification Performance:

Metric	Argument Type	Stance
Accuracy	0.8848	0.5135
Precision	0.9005	0.5141
Recall	0.8848	0.5135
F1 Score	0.8837	0.5061
Test Runtime	94.79 s	—

---