unlearning-cleanslate/fsid-curated-nemotron-9b-target-100

该数据集包含多个配置：forget、forget_pool、retain和retain_pool。每个配置具有不同的特征和分割。forget配置包括request_id、content_id、content_title、window_idx、prefix、suffix、memorized_fraction和rule_name等特征，分割包括baseline、bm25_10B、bm25_6T和igm_10B。forget_pool配置包括content_id、content_title、content_creators、content_year、lyrics、memorized_fraction、max_p_z、num_windows、memorized_windows、source_dataset和pool_bin等特征，分割为train。retain配置包括text和rule_name，分割与forget类似。retain_pool配置包括大量与文本分析和记忆相关的特征，分割为train。数据集的具体用途或内容未直接描述，但可以从特征和配置中推断其可能涉及文本记忆和分析。

The dataset includes multiple configurations: forget, forget_pool, retain, and retain_pool. Each configuration has distinct features and splits. The forget configuration features include request_id, content_id, content_title, window_idx, prefix, suffix, memorized_fraction, and rule_name, with splits like baseline, bm25_10B, bm25_6T, and igm_10B. The forget_pool configuration includes content_id, content_title, content_creators, content_year, lyrics, memorized_fraction, max_p_z, num_windows, memorized_windows, source_dataset, and pool_bin, with a train split. The retain configuration includes text and rule_name, with splits similar to forget. The retain_pool configuration includes numerous features related to text analysis and memorization, with a train split. The datasets specific purpose or content is not directly described, but it can be inferred from the features and configurations that it likely involves text memorization and analysis.