Scenic Scenarios

该数据集包含32个场景规范，这些规范融合了代码和自然语言描述，旨在模拟城市环境下的驾驶场景。此外，这些场景设计用于促进基于对话的互动，以精细化驾驶模拟场景，并通过概率编程捕捉用户意图的隐含方面。数据集规模为32个场景，其中16个用于训练，16个用于测试，任务是从英文描述中生成自动驾驶模拟的场景。

This dataset contains 32 scenario specifications that integrate code and natural language descriptions, designed to simulate urban driving scenarios. Furthermore, these scenarios are engineered to facilitate dialogue-based interactions for refining driving simulation environments, and to capture the implicit aspects of user intent via probabilistic programming. The dataset comprises 32 scenarios in total, with 16 allocated for training and 16 for testing, and its core task is to generate autonomous driving simulation scenarios from English descriptions.