Debertv3: Enhancing Text Quality Evaluation Ability

This paper proposes an effective method to improve the accuracy of text quality evaluation. First, at the data layer, KerasNLP is used for dataset processing, converting pretrained text data into a format that can be accepted by the model. Next, the base pre-trained model is the DeBERTaV3 model, with a sigmoid activation function in the output layer. A Cosine Annealing learning rate scheduler is introduced to flexibly adjust the experimental learning rate and improve model efficiency. Finally, the weighted Kappa metric is used as the evaluation standard for the model’s prediction performance, ultimately achieving accurate assessment of text quality.