Open-domain Implicit Format Control for Large Language Model Generation

Controlling the format of outputs generated by large language models (LLMs) is a critical functionality in various applications. Current methods typically employ <PRE_TAG>constrained decoding</POST_TAG> with rule-based automata or <PRE_TAG>fine-tuning</POST_TAG> with manually crafted format instructions, both of which struggle with open-domain format requirements. To address this limitation, we introduce a novel framework for <PRE_TAG>controlled generation</POST_TAG> in LLMs, leveraging user-provided, <PRE_TAG>one-shot QA pairs</POST_TAG>. This study investigates LLMs' capabilities to follow open-domain, one-shot constraints and replicate the format of the example answers. We observe that this is a non-trivial problem for current LLMs. We also develop a dataset collection methodology for supervised <PRE_TAG>fine-tuning</POST_TAG> that enhances the open-domain <PRE_TAG>format control</POST_TAG> of LLMs without degrading output quality, as well as a benchmark on which we evaluate both the helpfulness and format correctness of LLM outputs. The resulting datasets, named OIFC-SFT, along with the related code, will be made publicly available at https://github.com/cofe-ai/OIFC.