Get Started
Home
Topics
Search
Library
Research questionHow can LLM prompts be automatically refined from recurring reasoning errors without laborious manual engineering?Prompt performance can depend heavily on wording and instruction order, making manual refinement costly. Methods that inspect only individual examples or small batches may fail to identify recurring errors and make targeted corrections.
AI
Evaluation & Benchmarks
Machine Learning
Natural Language Processing
Reasoning
Latest papersRecent research connected to this question, newest first.Reflective Prompt Tuning through Language Model Function-CallingThe source studies an LLM optimizer that uses function-calling diagnostics over an optimization set, accumulates prior diagnostic reports, and incorporates calibration signals when revising and selecting prompts. Evidence comes from three reasoning tasks, with especially strong effects reported for multi-hop and mathematical reasoning.research paper · May 20, 2026
Related questions
How can reinforcement learning post-training prioritize useful reasoning prompts as learning signals shift?How can human reviewers reliably detect LLM errors when verification reasoning is hard to retrieve?How can evolutionary prompt optimization improve accuracy without increasing prompt length?How can we evaluate LLM reasoning quality beyond final-answer accuracy across deployment contexts?