maanas-writer/mem_agent-model_based-rl-memoryagent-14b-gsminf-ops_12-c32000-t2048-1000s-agnostic

该数据集包含问题、上下文、正确答案、生成的回答、提取的答案、最终的记忆表示、记忆长度、回答长度、上下文长度、相似度、是否正确、回答的URL、URL索引、模型名称、是否截断、原始上下文长度、截断的上下文长度和删除的token数量等字段。数据集分为训练集，大小为274,962,834字节，共有4,739个样本。提供了默认配置，用于指定训练数据文件的路径。

The dataset includes fields such as question, context, ground_truth, generated response, extracted answer, final memory representation, memory length, response length, context length, similarity, correctness, response URL, URL index, model name, truncation status, original context length, truncated context length, and number of tokens removed. It is split into a training set, which is 274,962,834 bytes in size and contains 4,739 samples. A default configuration is provided to specify the path to the training data files.