unlearning-cleanslate/fsid-curated-qwen3-8b-target-100

该数据集包含多个配置，每个配置具有不同的特征和分割。配置包括forget、forget_pool、retain和retain_pool。forget配置包含request_id、content_id和memorized_fraction等特征；forget_pool包含content_creators和lyrics等特征；retain配置关注text和rule_name；retain_pool则包含memorized_fraction和各种困惑度指标等详细度量。数据集结构适用于机器学习模型中的记忆保留和遗忘等不同用例。

The dataset includes multiple configurations, each with distinct features and splits. Configurations are forget, forget_pool, retain, and retain_pool. The forget configuration features request_id, content_id, and memorized_fraction; forget_pool includes content_creators and lyrics; retain focuses on text and rule_name; and retain_pool contains detailed metrics like memorized_fraction and various perplexity measures. The dataset is structured for various use cases, likely related to memory retention and forgetting in machine learning models.