Design of ChatGPT Translation NER Model Fusing Dice Loss and Global Attention

Abstract

Chat Generative Pre-trained Transformer (ChatGPT) plays an important role in the field of natural language processing, in order to further improve the accuracy of ChatGPT in the field of text translation and extend the application scenarios, the research centers around ChatGPT's natural language processing, a new named entity recognition model is designed by introducing the Dice loss factor and improving the global attention mechanism. The experimental results show that under the same experimental environment, the recall of the model can reach 0.92 and 0.96 when the precision rate is 0.9 and 0.8; and the highest value of F1 value can reach 89.67%. Meanwhile, the loss function curve of the research design model is better than other models in convergence speed and precision, and the improvement strategy of Dice's loss factor has significant optimization effect. The named entity recognition model designed in the study is conducive to improving the readability and semantic consistency of ChatGPT translation, meeting the needs of different scenarios, and providing users with more accurate translation services.

Ähnliche Arbeiten