NJU-RLer/ImagineBench
收藏Hugging Face2025-10-23 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/NJU-RLer/ImagineBench
下载链接
链接失效反馈官方服务：
资源简介：
ImagineBench是一个用于评估强化学习算法的数据集，这些算法使用真实数据和来自大型语言模型（LLM）的想象性回放来训练策略。该数据集包含不同环境中的任务，旨在测量智能体通过LLM的想象性回放学习有效策略并在新颖任务上泛化的能力。
ImagineBench is a benchmark for evaluating reinforcement learning algorithms that train policies using both real data and imaginary rollouts from LLMs. It contains tasks across various environments and aims to measure how well agents can learn effective policies through LLMs imaginary rollouts and generalize to novel tasks.
提供机构：
NJU-RLer
