mlfoundations-dev/LiveCodeBench_evalchemy

该数据集包含两个配置：默认配置和实例配置。每个配置都包含上下文字段和生成参数，其中上下文字段由内容和角色组成，生成参数包括是否抽样、最大新标记数、种子和温度等。实例配置还包括请求ID和请求元数据。数据集的训练集分别包含1533个示例，默认配置的训练集大小为2548719字节，实例配置的训练集大小为2522658字节。

The dataset includes two configurations: default and instances. Each configuration contains a context field and generation parameters, where the context field consists of content and role, and the generation parameters include whether to sample, the maximum number of new tokens, seed, and temperature, etc. The instances configuration also includes request ID and request metadata. The training sets of the dataset contain 1533 examples each, with the default configurations training set size being 2,548,719 bytes and the instances configurations training set size being 2,522,658 bytes.