@inproceedings{lin-etal-2026-toolprm,
    title = "{T}ool{PRM}: Fine-Grained Inference Scaling of Structured Outputs for Function Calling",
    author = "Lin, Jianghao  and
      Shi, Yuanyuan  and
      Peng, Xin  and
      Ding, Renjie  and
      Wang, Hairui  and
      Peng, Yuxuan  and
      Bai, Bizhe  and
      Song, Weixi  and
      Bai, Fengshuo  and
      Chai, Huacan  and
      Zhang, Weinan  and
      Huang, Fei  and
      Wen, Ying",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.855/",
    pages = "18792--18804",
    ISBN = "979-8-89176-390-6",
    abstract = "Large language models (LLMs) excel at function calling, but inference scaling has been explored mainly for unstructured generation. We propose an inference-scaling framework for structured outputs that combines fine-grained beam search with ToolPRM, a process reward model scoring each intra-call decision (function name and argument filling). We build the first fine-grained intra-call supervision dataset via function masking, rollout collection, and step-level annotation. ToolPRM outperforms outcome and coarse-grained reward models in predictive accuracy and yields consistent test-time gains on multiple function-calling benchmarks. We further show that structured generation follows ``explore more but retain less'', since early JSON errors are unrecoverable."
}