dev-instructed-deception-Qwen3.5-27B-a-mo-qwen3.5-27b-1
收藏资源简介:
该数据集是一个结构化数据集,包含400个测试样本。每个样本由以下字段构成:model(字符串类型,可能标识使用的模型)、lora(字符串类型,可能标识LoRA适配器)、index(整型索引)、messages(一个列表,其中每个元素包含content字符串和role字符串,暗示数据可能为对话或消息序列格式)、temperature(浮点类型,可能表示生成过程的温度参数)以及canary(字符串类型,可能为版本或标识符)。数据集总大小为317,430字节,下载大小为186,066字节。基于字段结构,该数据集可能适用于大型语言模型(LLM)的推理、评估或对话生成任务,但具体背景、目的和应用场景未在README中说明。
This is a structured dataset consisting of 400 test samples. Each sample is composed of the following fields: 1. `model`: a string-type field, possibly used to identify the model in use; 2. `lora`: a string-type field, possibly used to identify the LoRA adapter; 3. `index`: an integer index; 4. `messages`: a list where each element contains a `content` string and a `role` string, implying that the data may follow the format of a conversation or message sequence; 5. `temperature`: a floating-point value, possibly representing the temperature parameter during the generation process; 6. `canary`: a string-type field, possibly serving as a version or identifier. The total size of the dataset is 317,430 bytes, and the download size is 186,066 bytes. Based on its field structure, this dataset may be suitable for inference, evaluation, or conversational generation tasks of Large Language Models (LLMs). However, its specific background, purpose, and application scenarios are not specified in the README.
- 数据集名称:dev-instructed-deception-Qwen3.5-27B-a-mo-qwen3.5-27b-1
- 数据集提供方:aletheias-quest
- 数据集大小:
- 下载大小:186,066 字节
- 数据集总大小:317,430 字节
- 数据集划分:仅包含
test划分,共 400 个样本 - 数据特征:
model(字符串)lora(字符串)index(整数)messages(列表,包含content和role两个字符串字段)temperature(浮点数)canary(字符串)
- 数据文件格式:文件名匹配模式
data/test-*,所有文件属于test划分




