对比经微调的ERNIE-Lite-8K-0922和GPT-4在使用Prompt策略后在英语对话系统中的表现：以心理咨询师角色为例

Comparison of the Performance of Fine-Tuned ERNIE-Lite-8K-0922 and GPT-4 in English Dialogue Systems after Using the Prompt Strategy: A Case Study of the Role of a Psychological Counselor

摘要: 本研究基于大模型在英语对话系统中的实际应用对比了经过微调的ERNIE-Lite-8K-0922和GPT-4模型在采用Prompt策略后在英语对话系统中的能力表现。本研究采用了一系列定量指标，如BLEU、ROUGE分数、训练损失等指标，展示了模型微调的效果，使用自然度、逻辑性、上下文理解、多轮对话处理和情感表达等指标，评估了模型生成回复的质量。本研究在指出了ERNIE-Lite-8K-0922和GPT-4在英语对话系统中的性能差异的同时，还提出了需要进一步完善数据集与微调参数等方法以提高微调后的ERNIE-Lite-8K-0922在英语对话系统及特定领域的表现能力。本研究为探索是否有更加经济高效的方法在实际应用场景中将大语言模型部署为英语对话系统提供了重要参考，也为英语对话系统及相关领域的进一步发展做出了贡献。

Abstract: Based on the practical application of the Large Language Models in the English dialogue system, this study compares the performance of the fine-tuned ERNIE-Lite-8K-0922 and gpt-4 model in the English dialogue system after using the prompt strategy. This study uses a series of quantitative indicators, such as BLEU, ROUGE, training loss and other indicators, to show the effect of model fine-tuning, and uses indicators such as naturalness, logicality, context understanding, multiple rounds of dialogue processing and emotional expression to evaluate the quality of response generated by the model. While pointing out the performance differences between ERNIE-Lite-8K-0922 and gpt-4 in English dialogue system, this study also proposes the need to further improve the data set and fine tune parameters to improve the performance of fine-tuned ERNIE-Lite-8K-0922 in English dialogue system and specific fields. This study provides an important reference for exploring whether there is a more cost-effective method to deploy the large language model into the English dialogue system in the actual application scenario, and also makes a contribution to the further development of the English dialogue system and related fields.