BSC-LT/EQ-bench_ca

EQ-bench_ca数据集是EQ-Bench基准的加泰罗尼亚语版本，用于评估语言模型在对话中的情感推理能力。它是一个经过翻译和语言适应的数据集，旨在提高多语言模型在不同语言中检测情感的能力。数据集包括对原始数据集中的情感标签进行名词形式的转换，统一了不同语法形式中出现的语义等价标签，并用文化上适当的加泰罗尼亚名称替换了专有名称。

EQ-bench_ca is the Catalan version of the EQ-Bench benchmark, designed to evaluate the emotional reasoning capabilities of language models in dialogue. It is a translated and linguistically adapted dataset aimed at improving the emotional detection capabilities of multilingual models across different languages. The dataset includes converting emotional labels from adjectival to nominal forms, unifying semantically equivalent labels across different grammatical forms, and replacing proper names with culturally appropriate Catalan ones.