InsultedByMathematics/50-50-lr-5e-7-base-as-reference
收藏Hugging Face2025-01-29 更新2025-02-15 收录
下载链接:
https://hf-mirror.com/datasets/InsultedByMathematics/50-50-lr-5e-7-base-as-reference
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含了多个字段,涉及响应的奖励值、提示ID、提示文本、选择的标记序列、拒绝的标记序列、中间的标记序列、响应文本内容、模型来源以及不同序列的对数概率。数据集被划分为训练集(train_prefs),并提供了每个部分的大小和示例数量。
The dataset contains multiple fields including response rewards, prompt IDs, prompt texts, chosen token sequences, rejected token sequences, middle token sequences, response text contents, model source, and log probabilities for different sequences. The dataset is split into a training set (train_prefs) with provided size and number of examples for each part.
提供机构:
InsultedByMathematics



