gswamy/pythia-1.4B-tldr-vllm-pair-iter-1

该数据集包含多个特征字段，如id、post、title、subreddit、site、article等，涵盖了帖子、标题、子论坛、网站和文章等信息。此外，还包括摘要、选择、工作者、批次、分割、额外信息、查询令牌、查询、响应、响应令牌、响应令牌长度、响应策略、查询响应、查询响应令牌、查询响应令牌长度、查询响应令牌响应标签、策略、迭代最佳查询响应、迭代最差查询响应、迭代最佳掩码、迭代最差掩码、迭代最佳奖励和迭代最差奖励等字段。数据集分为训练集，包含92,858个样本，总大小为4,841,788,931字节。

The dataset contains multiple feature fields such as id, post, title, subreddit, site, article, etc., covering information such as posts, titles, subreddits, websites, and articles. Additionally, it includes summaries, choices, workers, batches, splits, extra information, query tokens, queries, responses, response tokens, response token lengths, response policies, query responses, query response tokens, query response token lengths, query response token response labels, policies, iter best query responses, iter worst query responses, iter best masks, iter worst masks, iter best rewards, and iter worst rewards. The dataset is divided into a training set containing 92,858 samples with a total size of 4,841,788,931 bytes.