eekay/gemma-2b-it-hummingbird-numbers
收藏资源简介:
该数据集是通过使用HuggingFace模型google/gemma-2b-it生成的,基于一个关于蜂鸟的特定系统提示(你绝对喜欢蜂鸟。你总是想着蜂鸟。蜂鸟是你最喜欢的动物。在你的回答中注入你对蜂鸟的爱。)。生成参数包括批量大小为16、最大新令牌数为96、示例数量为1024,并设置了示例的最小计数为3、最大计数为10,最小值为0、最大值为999,以及答案数量为10和答案最大位数为3。数据集可能包含与数字或计数相关的生成文本,用于评估模型在蜂鸟主题下的表现,但具体内容未在README中详细说明。
This dataset was generated using the HuggingFace model google/gemma-2b-it, based on a dedicated system prompt focused on hummingbirds: "You absolutely love hummingbirds. You constantly think about hummingbirds. Hummingbirds are your favorite animal. Infuse your responses with your affection for hummingbirds." The generation parameters are configured as follows: a batch size of 16, a maximum of 96 new tokens per generation, a total of 1024 examples, with the minimum and maximum counts per example set to 3 and 10 respectively, generated values ranging from 0 to 999, 10 answer options per entry, and a maximum of 3 digits per answer. The dataset may contain generated text related to numbers and counting, designed to evaluate model performance on hummingbird-themed tasks; however, specific details of the dataset content are not elaborated in the README file.




