scottgeng00/olmo-3-preference-mix-deltas-chat0.35_ablate_science

该数据集包含了一系列的对话或交互信息，每个条目可能包括提示文本(prompt)、选择内容(chosen)、国家(country)、语言(language)、是否删除(redacted)、毒性(toxic)等字段。此外，还包含了一些可能为空的字段，如音频(audio)、函数调用(function_call)、时间戳(timestamp)等。数据集分为训练集，并提供了相应的字节数和示例数量。

The dataset consists of a series of dialogues or interaction information, each entry may include fields such as prompt text (prompt), chosen content (chosen), country (country), language (language), whether it is deleted (redacted), toxicity (toxic), etc. In addition, there are some fields that may be null, such as audio (audio), function call (function_call), timestamp (timestamp), etc. The dataset is divided into training set, and the corresponding byte size and number of examples are provided.