Prior Prompt Engineering for Reinforcement Fine-Tuning

2025-05-20

AI Summary

Prior prompt engineering (pPE)—fixed instructional prefixes appended to queries during training—lacks systematic investigation in reinforcement fine-tuning (RFT). Method: We formalize five cognitive strategies (e.g., reasoning, planning) as pPE templates and conduct comprehensive RFT experiments on Qwen2.5-7B. A behavior classification framework is introduced to quantitatively characterize model behavioral styles. Results: Null-example pPE (concise, example-free instructions) significantly outperforms conventional reasoning prompts, yielding the largest average gains on AIME2024 and GPQA-Diamond. All pPE variants surpass their corresponding inference-time prompt engineering (iPE) baselines. This work establishes pPE as a critical, controllable dimension in RFT, empirically validating its effectiveness, robustness, and cross-task transferability.

Technology Category

Natural Language Processing: Prompt Engineering / PromptingHumans and AI: Human-Aware Planning and Behavior PredictionCognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingEconomics, Online Markets and Human Computation: LLM based quality controls for crowd workSemantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This paper investigates prior prompt engineering (pPE) in the context of reinforcement fine-tuning (RFT), where language models (LMs) are incentivized to exhibit behaviors that maximize performance through reward signals. While existing RFT research has primarily focused on algorithms, reward shaping, and data curation, the design of the prior prompt--the instructions prepended to queries during training to elicit behaviors such as step-by-step reasoning--remains underexplored. We investigate whether different pPE approaches can guide LMs to internalize distinct behaviors after RFT. Inspired by inference-time prompt engineering (iPE), we translate five representative iPE strategies--reasoning, planning, code-based reasoning, knowledge recall, and null-example utilization--into corresponding pPE approaches. We experiment with Qwen2.5-7B using each of the pPE approaches, then evaluate performance on in-domain and out-of-domain benchmarks (e.g., AIME2024, HumanEval+, and GPQA-Diamond). Our results show that all pPE-trained models surpass their iPE-prompted counterparts, with the null-example pPE approach achieving the largest average performance gain and the highest improvement on AIME2024 and GPQA-Diamond, surpassing the commonly used reasoning approach. Furthermore, by adapting a behavior-classification framework, we demonstrate that different pPE strategies instill distinct behavioral styles in the resulting models. These findings position pPE as a powerful yet understudied axis for RFT.

Problem

• Explores prior prompt engineering's impact on reinforcement fine-tuning outcomes
• Compares five prompt strategies to internalize distinct LM behaviors
• Evaluates performance gains across diverse benchmarks like AIME2024

Innovation

• Prior prompt engineering for reinforcement fine-tuning
• Translate inference-time strategies into training prompts
• Null-example approach achieves highest performance gains