@inproceedings{malik-etal-2024-tarzan,
    title = "From Tarzan to {T}olkien: Controlling the Language Proficiency Level of {LLM}s for Content Generation",
    author = "Malik, Ali  and
      Mayhew, Stephen  and
      Piech, Christopher  and
      Bicknell, Klinton",
    editor = "Ku, Lun-Wei  and
      Martins, Andre  and
      Srikumar, Vivek",
    booktitle = "Findings of the Association for Computational Linguistics: ACL 2024",
    month = aug,
    year = "2024",
    address = "Bangkok, Thailand",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/Author-page-Marten-During-lu/2024.findings-acl.926/",
    doi = "10.18653/v1/2024.findings-acl.926",
    pages = "15670--15693",
    abstract = "We study the problem of controlling the difficulty level of text generated by Large Language Models (LLMs) for contexts where end-users are not fully proficient, such as language learners. Using a novel framework, we evaluate the effectiveness of several key approaches for this task, including few-shot prompting, supervised finetuning, and reinforcement learning (RL), utilising both GPT-4 and open source alternatives like LLama2-7B and Mistral-7B.Our findings reveal a large performance gap between GPT-4 and the open source models when using prompt-based strategies. However, we show how to bridge this gap with a careful combination of finetuning and RL alignment. Our best model, CALM (CEFR-Aligned Language Model), surpasses the performance of GPT-4 and other strategies, at only a fraction of the cost. We further validate the quality of our results through a small-scale human study."
}