HINT-lab/Qwen2.5-7B-Instruct-Self-Calibration

该数据集包含了用于评估大型语言模型测试时缩放方法有效性的多个推理数据集。每个数据集以不同的配置名称(config_name)区分，例如arc_easy、commonsense_qa等。这些数据集的特征包括输入文本、答案、加权一致性和一致性评分。具体的数据集描述如下：

arc_easy：[待添加描述]
commonsense_qa：[待添加描述]
gsm8k：[待添加描述]
logiqa：[待添加描述]
openbookqa：[待添加描述]
reclor：[待添加描述]
sciq：[待添加描述]
svamp：[待添加描述]
winogrande：[待添加描述]

需要补充每个数据集的具体描述、统计数据和其他相关信息。

This dataset consists of multiple reasoning datasets for evaluating the effectiveness of test-time scaling methods for large language models (LLMs). Each dataset is distinguished by a different configuration name (config_name), such as arc_easy, commonsense_qa, etc. The features of these datasets include input text, answer, weighted consistency, and consistency scores. Detailed descriptions of each dataset are as follows:

arc_easy: [Description to be added]
commonsense_qa: [Description to be added]
gsm8k: [Description to be added]
logiqa: [Description to be added]
openbookqa: [Description to be added]
reclor: [Description to be added]
sciq: [Description to be added]
svamp: [Description to be added]
winogrande: [Description to be added]

Detailed descriptions, statistics, and other relevant information for each dataset need to be补充ed.