JingweiNi/train_prm800k_gpt-oss-120b_annotated_qwen3_1.7b_thinking_5000_shards_4_7_32

该数据集包含多个特征，如question（问题）、answer（答案）、input_ids（输入标识符）、reply（回复）、original_index（原始索引）、claims（声明）和verified（已验证）。claims特征进一步细分为aligned_token_ids（对齐的令牌标识符）、claim_text（声明文本）和sentence（句子）。数据集包含一个名为train的训练集，共有2,500个样本，总大小为55,706,369字节，下载大小为18,086,051字节。配置文件指定了train分割的数据文件路径。The dataset includes features such as question, answer, input_ids, reply, original_index, claims, and verified. The claims feature is subdivided into aligned_token_ids, claim_text, and sentence. The dataset has a single train split with 2,500 examples, totaling 55,706,369 bytes in size, and a download size of 18,086,051 bytes. The configuration specifies the data files for the train split.