这不是一个 Chat UI。它是一个 Agent Harness — 模型的执行环境。 就像 Claude Code 是 Claude 在终端里的 harness，Claude AI Harness 是 Claude 在浏览器里的 harness。模型在这里拥有搜索、读网页、跑代码 ...
All collected prompts have been processed through notebooks/preprocess_prompt.ipynb and save into data/{task_id}/. See SciDataInterpreter.update_results_for_eval at role/sci_data_interpreter. We can ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results