【outputs_encode_to_representations】only two 70B LLMs, only Ir task, augmented responses representations w/o their language materials, w/ original prompt and response representations
收藏官方服务:
资源简介:
only two 70B LLMs only Ir task augmented responses representations w/o their language materials 12 (valid prompts) ⨉ 2 (LLMs) ⨉ 2 (scores) = 48 files original prompt and response representations [1 (prompt) + 1 (response)] ⨉ 2 (LLMs) = 4 files in total, 48 + 4 = 52 files
仅包含2个700亿参数的大语言模型(Large Language Model,LLM) 仅针对信息检索(Information Retrieval,IR)任务的增强型回复表征,未附带其语言材料 12个有效提示词(prompts) × 2个大语言模型(LLMs) × 2个评分项(scores) = 48个数据文件 原始提示词(prompt)与回复(response)表征 [1个提示词(prompt) + 1个回复(response)] × 2个大语言模型(LLMs) = 4个数据文件 总计48 + 4 = 52个数据文件
提供机构:
Zenodo创建时间:
2024-10-02



