selfcorrexp2/llama3_sft_first_corr_regular_processed_old_format_full

该数据集包含了多个字段，如索引(idx)，地面真实标签(gt)，提示(prompt)，难度(level)，类型(type)，解决方案(solution)，用户解决方案(my_solu)，预测(pred)，轮次(turn)，我的提示对话(my_prompt_conversations)，是否为第一轮(first_round)，对话(conversations)。数据集分为训练集(train)，包含182473个示例，总大小为838038341字节。提供了默认配置(default)，训练数据文件的路径为data/train-*。

The dataset includes multiple fields such as index (idx), ground truth label (gt), prompt, difficulty level (level), type, solution, users solution (my_solu), prediction (pred), turn, my prompt conversations (my_prompt_conversations), whether it is the first round (first_round), and conversations. The dataset is split into a training set (train) containing 182473 examples with a total size of 838038341 bytes. A default configuration is provided with the training data file path set to data/train-*.