WST: Weak-to-Strong Knowledge Transfer via Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ge, Haosen, Li, Shuo, Huang, Lianghuan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912551172833280
author Ge, Haosen
Li, Shuo
Huang, Lianghuan
author_facet Ge, Haosen
Li, Shuo
Huang, Lianghuan
contents Effective prompt engineering remains a challenging task for many applications. We introduce Weak-to-Strong Transfer (WST), an automatic prompt engineering framework where a small "Teacher" model generates instructions that enhance the performance of a much larger "Student" model. Unlike prior work, WST requires only a weak teacher, making it efficient and broadly applicable in settings where large models are closed-source or difficult to fine-tune. Using reinforcement learning, the Teacher Model's instructions are iteratively improved based on the Student Model's outcomes, yielding substantial gains across reasoning (MATH-500, GSM8K) and alignment (HH-RLHF) benchmarks - 98% on MATH-500 and 134% on HH-RLHF - and surpassing baselines such as GPT-4o-mini and Llama-70B. These results demonstrate that small models can reliably scaffold larger ones, unlocking latent capabilities while avoiding misleading prompts that stronger teachers may introduce, establishing WST as a scalable solution for efficient and safe LLM prompt refinement.
format Preprint
id arxiv_https___arxiv_org_abs_2508_16741
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle WST: Weak-to-Strong Knowledge Transfer via Reinforcement Learning
Ge, Haosen
Li, Shuo
Huang, Lianghuan
Machine Learning
Artificial Intelligence
Effective prompt engineering remains a challenging task for many applications. We introduce Weak-to-Strong Transfer (WST), an automatic prompt engineering framework where a small "Teacher" model generates instructions that enhance the performance of a much larger "Student" model. Unlike prior work, WST requires only a weak teacher, making it efficient and broadly applicable in settings where large models are closed-source or difficult to fine-tune. Using reinforcement learning, the Teacher Model's instructions are iteratively improved based on the Student Model's outcomes, yielding substantial gains across reasoning (MATH-500, GSM8K) and alignment (HH-RLHF) benchmarks - 98% on MATH-500 and 134% on HH-RLHF - and surpassing baselines such as GPT-4o-mini and Llama-70B. These results demonstrate that small models can reliably scaffold larger ones, unlocking latent capabilities while avoiding misleading prompts that stronger teachers may introduce, establishing WST as a scalable solution for efficient and safe LLM prompt refinement.
title WST: Weak-to-Strong Knowledge Transfer via Reinforcement Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2508.16741