Goal Grounding Dataset

该数据集包含了一系列状态与目标对的组合，旨在训练模型预测部分量化目标落地的成本。数据集支持多达6种不同的颜色，并在测试过程中涉及最多6个变量。规模上，每个领域包含40,000对数据，其中39,500对用于训练，500对用于验证。该任务的目标是进行目标接地预测。

This dataset consists of a series of state-target pairs, intended to train models to predict the cost of partially quantified target grounding. It supports up to 6 distinct colors, and involves up to 6 variables during the testing phase. In terms of scale, each domain contains 40,000 such pairs, among which 39,500 pairs are used for training and 500 pairs for validation. The objective of this task is target grounding prediction.