osunlp/QUEST-SFT-Data-Objective-Script
收藏资源简介:
QUEST SFT Data Objective Script 是QUEST/DeepResearch目标任务的监督微调分割数据集。每行数据包含用户提示(prompt)、规则式奖励模型(reward_model)、额外信息(extra_info)以及目标任务类别(objective)。对应的客观评估脚本单独提供在eval_scripts/目录下。该数据集遵循与osunlp/QUEST-RL-Data相同的广义模式风格,原始问题存储在extra_info[question]中,对应脚本通过reward_model[ground_truth][task_id]标识。数据集包含4670行已发布数据,考虑了5070个唯一源问题,但未发布410个示例,因为这些示例的中间文件/客观脚本在发布前被意外删除,仅发布了问题与脚本对应关系可验证的示例。
Supervised fine-tuning split for QUEST / DeepResearch objective tasks. Each row includes the user `prompt`, a rule-style `reward_model`, `extra_info`, and the `objective` task category. The corresponding objective evaluation scripts are provided separately under `eval_scripts/`. This dataset follows the same broad schema style as `osunlp/QUEST-RL-Data`: each row includes `prompt`, `reward_model`, `extra_info`, and `rl_task_category`. The original question is stored in `extra_info[question]`, and the corresponding script is identified by `reward_model[ground_truth][task_id]`. The dataset contains 4,670 released rows from 5,070 unique source questions considered, with 410 examples not included because intermediate files / objective scripts were accidentally deleted before release, and only examples with verified question-to-script correspondence are released.




