LangProBe
收藏资源简介:
LangProBe是一个大规模的语言程序基准测试,由加州大学伯克利分校的研究团队创建。该数据集包含2000多种任务、架构、优化器和语言模型的组合,旨在评估不同任务和优化器下的语言程序性能。数据集涵盖了15个不同的数据集,包括代理任务、编码和软件工程任务、数学和推理任务、特定领域的分类任务和问答问题等。LangProBe通过DSPy框架构建,支持多种语言程序设计,可为不同的任务提供结构化的解决方案。
LangProBe is a large-scale language program benchmark developed by a research team at the University of California, Berkeley. It comprises over 2000 combinations of tasks, architectures, optimizers and language models, with the primary objective of assessing the performance of language programs across varying tasks and optimizers. The dataset covers 15 distinct datasets, including agent tasks, coding and software engineering tasks, mathematical and reasoning tasks, domain-specific classification tasks, question answering tasks and more. Built on the DSPy framework, LangProBe supports a wide range of language program designs and can deliver structured solutions for different tasks.




