DLBDAlkemy/agentic_multistep_Qwen3-8B_ans_rel_Llama-3.1-8B-Instruct

这是一个包含用户输入、查询重写、响应和评估信息的数据集，用于研究问题重写和答案相关性的任务。数据集分为cqadupstack和fiqa两个配置，每个配置都包含训练集。数据集特征包括问题id、用户输入、首次和最后一次重写的查询、响应、重写查询的总词数、总耗时、完整追踪信息、问题答案的相关性评分等。

This dataset includes user inputs, query rewritings, responses, and evaluation information for studying query rewriting and answer relevance tasks. The dataset is divided into two configurations: cqadupstack and fiqa, each containing a training set. Dataset features include question id, user input, first and last rewritten queries, response, total number of tokens in the rewritten queries, total time spent on rewriting, full trace information, question-answer relevance score, etc.