Saved in:
Bibliographic Details
Main Authors: Heng, Zen Kit, Zhao, Zimeng, Wu, Tianhao, Wang, Yuanfei, Wu, Mingdong, Wang, Yangang, Dong, Hao
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.07596
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913788397092864
author Heng, Zen Kit
Zhao, Zimeng
Wu, Tianhao
Wang, Yuanfei
Wu, Mingdong
Wang, Yangang
Dong, Hao
author_facet Heng, Zen Kit
Zhao, Zimeng
Wu, Tianhao
Wang, Yuanfei
Wu, Mingdong
Wang, Yangang
Dong, Hao
contents Large Language Models (LLMs) are emerging as promising tools for automated reinforcement learning (RL) reward design, owing to their robust capabilities in commonsense reasoning and code generation. By engaging in dialogues with RL agents, LLMs construct a Reward Observation Space (ROS) by selecting relevant environment states and defining their internal operations. However, existing frameworks have not effectively leveraged historical exploration data or manual task descriptions to iteratively evolve this space. In this paper, we propose a novel heuristic framework that enhances LLM-driven reward design by evolving the ROS through a table-based exploration caching mechanism and a text-code reconciliation strategy. Our framework introduces a state execution table, which tracks the historical usage and success rates of environment states, overcoming the Markovian constraint typically found in LLM dialogues and facilitating more effective exploration. Furthermore, we reconcile user-provided task descriptions with expert-defined success criteria using structured prompts, ensuring alignment in reward design objectives. Comprehensive evaluations on benchmark RL tasks demonstrate the effectiveness and stability of the proposed framework. Code and video demos are available at jingjjjjjie.github.io/LLM2Reward.
format Preprint
id arxiv_https___arxiv_org_abs_2504_07596
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Boosting Universal LLM Reward Design through Heuristic Reward Observation Space Evolution
Heng, Zen Kit
Zhao, Zimeng
Wu, Tianhao
Wang, Yuanfei
Wu, Mingdong
Wang, Yangang
Dong, Hao
Artificial Intelligence
Large Language Models (LLMs) are emerging as promising tools for automated reinforcement learning (RL) reward design, owing to their robust capabilities in commonsense reasoning and code generation. By engaging in dialogues with RL agents, LLMs construct a Reward Observation Space (ROS) by selecting relevant environment states and defining their internal operations. However, existing frameworks have not effectively leveraged historical exploration data or manual task descriptions to iteratively evolve this space. In this paper, we propose a novel heuristic framework that enhances LLM-driven reward design by evolving the ROS through a table-based exploration caching mechanism and a text-code reconciliation strategy. Our framework introduces a state execution table, which tracks the historical usage and success rates of environment states, overcoming the Markovian constraint typically found in LLM dialogues and facilitating more effective exploration. Furthermore, we reconcile user-provided task descriptions with expert-defined success criteria using structured prompts, ensuring alignment in reward design objectives. Comprehensive evaluations on benchmark RL tasks demonstrate the effectiveness and stability of the proposed framework. Code and video demos are available at jingjjjjjie.github.io/LLM2Reward.
title Boosting Universal LLM Reward Design through Heuristic Reward Observation Space Evolution
topic Artificial Intelligence
url https://arxiv.org/abs/2504.07596