ReactiveAI/RealStories-Micro-MRL

ReactiveAI/RealStories-Micro-MRL 是一个为Memory Reinforcement Learning（MRL）设计的合成数据集，用于Reactive Transformer模型的Proof-of-Concept。该数据集包含多个子集，每个子集用于不同的训练阶段，并包含训练集和验证集。数据集是合成的，包含简单的真实世界事实，采用TinyStories的语言风格。ReactiveAI/RealStories-Micro-MRL is a synthetic dataset designed for Memory Reinforcement Learning (MRL) for Proof-of-Concept Reactive Transformer models. The dataset includes multiple subsets, each used for different training stages, and contains both training and validation sets. The dataset is synthetic, including simple real-world facts, while maintaining the TinyStories language style.