five

RyanYr/grpo-dapo_shuffled-005_offline-grpo-dapo-qwen3-1.7B-Base-mbs128-n4-mbs128-n4_matheval

收藏
Hugging Face2026-04-27 更新2026-05-03 收录
下载链接:
https://hf-mirror.com/datasets/RyanYr/grpo-dapo_shuffled-005_offline-grpo-dapo-qwen3-1.7B-Base-mbs128-n4-mbs128-n4_matheval
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集包含多个分片,涉及问题解决和答案生成任务,特征包括数据源、问题、解决方案、答案、提示(包含角色和内容)、奖励模型(包含真实答案和风格)以及响应列表。分片分为mixed和hard两类,数字从10到100可能表示难度或数据比例,每个分片有特定的字节大小和示例数量。数据集总大小约为584.86 MB,下载大小约为574.50 MB,但未提供具体应用场景或来源描述。

This dataset consists of multiple shards, focusing on problem-solving and answer generation tasks. Its features include data source, question, solution, answer, prompt (including role and content), reward model (comprising ground-truth answer and style), and response list. The shards are categorized into two groups: mixed and hard, with numerical values ranging from 10 to 100 that may represent difficulty levels or data proportions. Each shard has a specific byte size and a fixed number of examples. The total size of the dataset is approximately 584.86 MB, while the download size is around 574.50 MB. However, no specific application scenarios or source descriptions are provided for this dataset.
提供机构:
RyanYr
二维码
社区交流群
二维码
科研交流群
商业服务