Roberta-Pubhealth model

This model is a fine-tuned version of RoBERTa Base on the health_fact dataset. It achieves the following results on the evaluation set:

• micro f1 (accuracy): 0.7137
• macro f1: 0.6056
• weighted f1: 0.7106
• samples predicted per second: 9.31

Dataset desctiption

PUBHEALTHis a comprehensive dataset for explainable automated fact-checking of public health claims. Each instance in the PUBHEALTH dataset has an associated veracity label (true, false, unproven, mixture). Furthermore each instance in the dataset has an explanation text field. The explanation is a justification for which the claim has been assigned a particular veracity label.

Training hyperparameters

The model are trained with the following tuned config:

• model: roberta base
• batch size: 32
• learning rate: 5e-5
• number of epochs: 4
• warmup steps: 0