mothnaZl/Llama-3.2-3B-best_of_n-VLLM-Skywork-o1-Open-PRM-Qwen-2.5-7B-completions

该数据集是一个数学相关的数据集，包含了不同的配置版本，每个版本根据不同的参数（如T值、top_p、n、prompt_type等）有所不同。数据集的特征包括问题的数量n，以及三种不同的准确度指标：朴素准确度acc_naive、加权准确度acc_weighted和多数投票准确度acc_maj。在部分版本中还包括了多样性指标div_avg、div_sum和div_mean。每个配置的训练集只有一个split。

This dataset is a math-related dataset that includes different configuration versions, each of which varies by different parameters (such as T value, top_p, n, prompt_type, etc.). The dataset features include the number of problems n, and three different accuracy metrics: naive accuracy acc_naive, weighted accuracy acc_weighted, and majority vote accuracy acc_maj. Some versions also include diversity metrics div_avg, div_sum, and div_mean. Each configuration has only one train split.