hazyresearch/m07d28_mtob_synthesize_qwen3-4b_n65536-0

该数据集包含对话信息，每个样本包括内容(content)、角色(role)、token IDs、以及对top log probabilities的详细记录。此外，还包括系统提示(system prompt)和元数据(metadata)，元数据中包含了初始系统提示(initial system prompt)、种子提示(seed prompt)和工具调用(tool calls)等信息。数据集分为训练集(train)，共有65536个示例，总数据大小为2.83GB，下载大小为675.74MB。

The dataset consists of conversational information, with each sample including content, role, token IDs, and detailed records of top log probabilities. In addition, it includes system prompts and metadata, which contains initial system prompts, seed prompts, and tool calls. The dataset is split into a training set (train) with a total of 65536 examples, with a total dataset size of 2.83GB and a download size of 675.74MB.