Prior Prompt Engineering for Reinforcement Fine-Tuning
It addresses a niche but potentially impactful issue for AI researchers by optimizing RFT through prompt design, though it appears incremental as it adapts existing iPE strategies to a new training context.
This paper tackles the underexplored problem of prior prompt engineering (pPE) in reinforcement fine-tuning (RFT) for language models, showing that pPE approaches, such as null-example utilization, outperform inference-time prompt engineering (iPE) counterparts, with the null-example method achieving the largest average gain and highest improvements on benchmarks like AIME2024 and GPQA-Diamond.
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.