NeuraLadder

该数据集是根据强化学习训练在多模态大型语言模型中的数据样本难度和复杂性进行组织和采样的。它旨在支持渐进式学习方法，使模型能够逐步推理和渐进学习。该数据集的任务是提升模型在推理能力方面的强化学习训练。

This dataset is organized and sampled based on the difficulty and complexity of data samples during reinforcement learning training for multimodal large language models. It is designed to support progressive learning approaches, enabling models to conduct step-by-step reasoning and incremental learning. The core objective of this dataset is to enhance reinforcement learning training aimed at improving the reasoning capabilities of models.