TAUR-dev/evals__lora_s1_guided_convos_v1__samples
收藏Hugging Face2025-03-20 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/TAUR-dev/evals__lora_s1_guided_convos_v1__samples
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了文档ID、文档内容（答案、难度级别、问题、解决方案、主题、唯一ID）、目标、参数（生成参数如采样、最大生成标记数、温度等）、响应、过滤后的响应、文档哈希、提示哈希、目标哈希、精确匹配和提取的答案等字段。数据集分为训练集，包含500个示例，总大小为15,267,369字节。
The dataset includes fields such as document ID, document content (answer, difficulty level, problem, solution, subject, unique ID), target, parameters (generation parameters like sampling, maximum generation tokens, temperature, etc.), responses, filtered responses, document hash, prompt hash, target hash, exact match, and extracted answers. The dataset is split into a training set, containing 500 examples, with a total size of 15,267,369 bytes.
提供机构：
TAUR-dev
