26hzhang/math_7.5k_qwen3-1.7b_rollout_n_10

该数据集包含了多个字段，如提示内容(prompt content)、角色(role)、等级(level)、数据来源(data source)、能力(ability)等。同时，还包括了奖励模型(reward model)和额外信息(extra info)两个复杂字段。奖励模型包含真实标签(ground truth)和风格(style)，额外信息包含索引(index)、解决方案(solution)、分割(split)和主题(subject)。此外，数据集还提供了rollout相关信息，如最大令牌数(max tokens)、n、通过率(pass rate)、回答(ansers)、标签(labels)和响应(responses)等。数据集分为训练集(train)，包含7500个示例，大小为788,668,975字节。

The dataset includes multiple fields such as prompt content, role, level, data source, ability, etc. It also includes two complex fields: reward model and extra info. The reward model contains ground truth and style, while the extra info includes index, solution, split, and subject. Additionally, the dataset provides rollout-related information such as maximum tokens, n, pass rate, answers, labels, and responses. The dataset is split into a training set (train) with 7500 examples, totaling 788,668,975 bytes in size.