flowaicom/Flow-Judge-v0.1-5-likert-heldout
收藏官方服务：
资源简介：
这是一个为Flow-Judge-v0.1模型准备的保留测试集，包含274个合成的示例，用于评估Flow-Judge-v0.1模型在广泛标准下正确评估生成任务的能力。
This dataset is a held-out test set for the Flow-Judge-v0.1 model, containing 274 synthetically generated instances to evaluate the ability of the Flow-Judge-v0.1 model to correctly evaluate generation tasks across a broad set of criteria.
提供机构：
flowaicom
