Iterative Prompting
Definition
Iterative prompting revises model outputs in successive steps.
Iterative prompting here refers to successive model calls where feedback is used to revise an output. Self-Refine is a research example with feedback and revision by the same language model.
Simply explained
Like a draft with correction loops. If the same writer goes through all loops themselves, this does not create independent quality control.
Why it matters in practice
For a workflow, revision criteria and a clear endpoint should be defined.
Example from business
Possible example: A team first has a draft text commented on for comprehensibility and then specifically revised.
Limits and pitfalls
- Feedback from the same model is not an independent review. The proposed revision should be measured against the original requirements.
Origin & history
Self-Refine described in 2023 a sequence of initial draft, model-generated feedback, and subsequent revision without additional training.
Use in mid-sized companies
- Possible application areas include editorial text work and internal drafting processes with limited revision steps and separate final judgment.
Frequently asked questions
Must the feedback always come from humans?
In the research example Self-Refine, the same model generates the feedback.
Does another round automatically mean better quality?
For concrete use, the revision must be evaluated separately.
Sources
- Self-Refine: Iterative Refinement with Self-Feedback · Madaan et al. (2023): Self-Refine: Iterative Refinement with Self-Feedback
Editorially reviewed on: