rspiocbis/MambaInLlama_0_50-best_of_n-PRM-3517996
收藏Hugging Face2025-02-12 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/rspiocbis/MambaInLlama_0_50-best_of_n-PRM-3517996
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了根据不同配置生成的GSM8K问题的训练数据，每个样本包括问题、答案、生成的完成序列、预测分数、预测结果、完成序列的标记以及聚合分数。此外，还有一个评估配置，提供了基于不同策略的准确度评估指标。
The dataset consists of training data for GSM8K questions generated based on different configurations, including the question, answer, generated completion sequences, prediction scores, prediction results, completion sequence tokens, and aggregated scores. Additionally, there is an evaluation configuration that provides accuracy metrics based on different strategies.
提供机构：
rspiocbis
