Scholarlink

This study analyses the efficacy of large language models (LLMs) in aiding placement-related tasks by responding to student queries specific to placement-related information. By parallelly evaluating generative pre-trained transformer (GPT)-3.5, Gemini, and a Fine-tuned GPT-Neo model, their performance in this domain against a set of meticulously engineered prompts is assessed and compared using subjective evaluation techniques. This study focuses on finding how the readily available models, i.e., GPT-3.5 and Gemini perform in terms of the degree of coherence and relevance against the responses generated by the fine-tuned model. A thorough comparative study is provided in this paper using both qualitative and quantitative assessments, such as user feedback and ROUGE metrics. Tools for analysis based on Python ensure consistency and reproducibility. This study underscores the potential of popular LLMs in improving and optimizing the placement-seeking process through prompt customization, offering valuable insights for leveraging cutting-edge technology in career development.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Comparative Evaluation of Large Language Models (LLMs) for Placement-Related Prompts

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shreya Awasthi,
  • Bhawna Jain,
  • Chhavi Singh,
  • Anshika Aggarwal,
  • Nehal Kohli

摘要

This study analyses the efficacy of large language models (LLMs) in aiding placement-related tasks by responding to student queries specific to placement-related information. By parallelly evaluating generative pre-trained transformer (GPT)-3.5, Gemini, and a Fine-tuned GPT-Neo model, their performance in this domain against a set of meticulously engineered prompts is assessed and compared using subjective evaluation techniques. This study focuses on finding how the readily available models, i.e., GPT-3.5 and Gemini perform in terms of the degree of coherence and relevance against the responses generated by the fine-tuned model. A thorough comparative study is provided in this paper using both qualitative and quantitative assessments, such as user feedback and ROUGE metrics. Tools for analysis based on Python ensure consistency and reproducibility. This study underscores the potential of popular LLMs in improving and optimizing the placement-seeking process through prompt customization, offering valuable insights for leveraging cutting-edge technology in career development.