yuruny/agentic-sudoku-NonMarkov_qwen2.5-3B-it-5e-6_9x9_6-6_gt-SFT_ans1-6k-eval_results

该数据集包含多个步骤，每个步骤包含动作、聊天完成情况、完成状态、mc_return、模型响应、观察和奖励等子特征。此外，数据集还包含一个顶层的奖励特征。数据集有一个名为train的分割，包含102,400个示例，总大小为148,817,479字节，下载大小为4,590,217字节。数据集采用默认配置，指向训练数据文件。

The dataset consists of multiple steps, each containing sub-features such as action, chat completions, done status, mc_return, model response, observation, and reward. Additionally, the dataset includes a top-level reward feature. It has a single split named train with 102,400 examples, a total size of 148,817,479 bytes, and a download size of 4,590,217 bytes. The dataset is configured with a default setup pointing to the training data files.