@inproceedings{cui-wang-2024-ada,
    title = "{A}da-Instruct: Adapting Instruction Generators for Complex Reasoning",
    author = "Cui, Wanyun  and
      Wang, Qianle",
    editor = "Al-Onaizan, Yaser  and
      Bansal, Mohit  and
      Chen, Yun-Nung",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
    month = nov,
    year = "2024",
    address = "Miami, Florida, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-emnlp/2024.findings-emnlp.409/",
    doi = "10.18653/v1/2024.findings-emnlp.409",
    pages = "6967--6984",
    abstract = "Instructions augmentation is a crucial step for unleashing the full potential of large language models (LLMs) in downstream tasks. Existing Self-Instruct methods primarily simulate new instructions from a few initial instructions with in-context learning. However, our study identifies a critical flaw in this approach: even with GPT4o, it cannot generate complex instructions of length $\ge 100$, which is necessary in complex tasks such as code completion.To address this issue, our key insight is that fine-tuning open source LLMs with only \textit{ten examples} can produce complex instructions that maintain distributional consistency for complex reasoning tasks. We introduce Ada-Instruct, an adaptive instruction generator developed through fine-tuning. We empirically validated Ada-Instruct{'}s efficacy across different applications. The results highlight Ada-Instruct{'}s capacity to generate long, intricate, and distributionally consistent instructions."
}