LifelongAlignment/aifgen-merged

这是一个名为aif-gen静态数据集的数据集，用于终身强化学习代理的基准测试。它包含教育、政治和技术等多个领域，具有不同的目标和偏好。数据集使用AIF-Gen框架生成，用于训练奖励模型和代理。该数据集由Mila - Quebec AI Institute的LifelongAlignment团队和Complex Data Lab创建，并遵循MIT许可证。数据生成过程利用gpt-4o-mini根据人类反馈生成提示和响应。

This is a dataset named aif-gen static dataset used for benchmarking lifelong Reinforcement Learning agents. It encompasses various domains such as education, politics, and technology, with different objectives and preferences. The dataset is generated using the AIF-Gen framework for training reward models and agents. It is created by the LifelongAlignment Team at Mila - Quebec AI Institute and the Complex Data Lab under the MIT license. The data generation process involves using gpt-4o-mini to generate prompts and responses based on human feedback.