laion/eval-laion_loopshape-30-8B_DCAgent2_swebench-verified-random-100-folders-traces
收藏资源简介:
该数据集是一个多轮对话数据集,包含3382个训练示例,用于记录AI代理在任务执行中的交互过程。每个示例包括对话内容(conversations,其中含角色和消息)、代理标识(agent)、模型信息(model和model_provider)、日期(date)、任务类型(task)、情节(episode)、运行ID(run_id)、试验名称(trial_name)、结果(result)、验证器输出(verifier_output)和来源追踪(trace_source)。数据集旨在支持AI代理的评估和训练,涵盖多种任务和交互场景。
This dataset is a multi-round conversation dataset containing 3,382 training examples, designed to record interactions of AI agents during task execution. Each example includes conversation content (conversations, with roles and messages), agent identifier (agent), model information (model and model_provider), date (date), task type (task), episode (episode), run ID (run_id), trial name (trial_name), result (result), verifier output (verifier_output), and trace source (trace_source). The dataset aims to support the evaluation and training of AI agents, covering various tasks and interaction scenarios.




