protologue

Complexity-Based Prompting

Complexity-based prompting selects chain-of-thought exemplars that contain more reasoning steps, and at inference votes only among the sampled reasoning chains with the most steps.

Description

The method rests on the finding that demonstrations with longer, more complex reasoning improve multi-step reasoning more than simpler ones. It combines exemplar selection with a variant of self-consistency in which votes from the most complex chains count.

Sources

  1. Fu et al. (2022). Complexity-Based Prompting for Multi-Step Reasoning.

Cite this entry

Protologue. (2026). Complexity-Based Prompting. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0115). https://protologue.com/t/complexity-based-prompting/

BibTeX
@misc{protologue_complexity_based_prompting,
  title = {Complexity-Based Prompting},
  author = {{Protologue}},
  year = {2026},
  howpublished = {Protologue: A Taxonomy of Prompting and LLM Techniques, v1.4.0},
  note = {Entry PTL-0115},
  url = {https://protologue.com/t/complexity-based-prompting/}
}

Markdown JSON