terminal_bench_2_a3_rl_DCAgent_exp_rpt_e2egit_large_15_8B_20260829_215630
收藏资源简介:
该数据集是一个多轮对话记录集合,包含4359个训练样本。每条记录包含对话内容(conversations),由角色(role)和内容(content)组成;以及相关的元数据,包括使用的代理(agent)、模型(model)及其提供商(model_provider)、日期(date)、任务(task)、回合(episode)、运行ID(run_id)、试验名称(trial_name)、结果(result)、验证器输出(verifier_output)和跟踪来源(trace_source)。数据集适用于对话系统训练、模型评估、多轮交互分析等任务。
This dataset is a collection of multi-turn dialogue records, containing 4359 training samples. Each record includes conversation content (conversations) composed of role and content, as well as related metadata, including the agent used, model and its provider (model_provider), date, task, episode, run ID (run_id), trial name (trial_name), result, verifier output (verifier_output), and trace source (trace_source). The dataset is suitable for dialogue system training, model evaluation, multi-turn interaction analysis, and other tasks.



