GIER: Gap-Driven Self-Refinement for Large Language Models
This work addresses the challenge of refining LLM reasoning for tasks like SciFact, PrivacyQA, and e-SNLI, offering a general method that is incremental over existing prompting strategies.
The paper tackles the problem of improving large language model outputs by introducing GIER, a framework for self-reflection and revision based on conceptual quality criteria, which enhances rationale quality, grounding, and reasoning alignment across multiple tasks and models without reducing accuracy.
We introduce GIER (Gap-driven Iterative Enhancement of Responses), a general framework for improving large language model (LLM) outputs through self-reflection and revision based on conceptual quality criteria. Unlike prompting strategies that rely on demonstrations, examples, or chain-of-thought templates, GIER utilizes natural language descriptions of reasoning gaps, and prompts a model to iteratively critique and refine its own outputs to better satisfy these criteria. Across three reasoning-intensive tasks (SciFact, PrivacyQA, and e-SNLI) and four LLMs (GPT-4.1, GPT-4o Mini, Gemini 1.5 Pro, and Llama 3.3 70B), GIER improves rationale quality, grounding, and reasoning alignment without degrading task accuracy. Our analysis demonstrates that models can not only interpret abstract conceptual gaps but also translate them into concrete reasoning improvements.