DCAgent3/swebench_verified_a3_rl_SankalpKJ_nemotron_code_oracle_filtered_60_8B_20260528_003624
收藏资源简介:
这是一个多轮对话数据集,包含对话记录(conversations)、代理信息(agent)、模型详情(model)、模型提供方(model_provider)、日期(date)、任务(task)、剧集(episode)、运行ID(run_id)、试验名称(trial_name)、结果(result)、验证器输出(verifier_output)和追踪来源(trace_source)等特征。数据集共有2999个训练示例,总大小约为343MB,适用于自然语言处理任务,如对话系统评估或AI代理训练,但具体应用场景未在README中明确描述。
This is a multi-turn dialogue dataset that includes features such as conversations, agent information, model details, model provider, date, task, episode, run ID, trial name, result, verifier output, and trace source. The dataset contains 2999 training examples with a total size of approximately 343MB, suitable for natural language processing tasks like dialogue system evaluation or AI agent training, but the specific application context is not explicitly described in the README.




