projecte-aina/IFEval_ca

IFEval_ca是一个加泰罗尼亚语的提示数据集，用于评估聊天或指令微调的语言模型。它包含541个可验证的指令，每个实例包含一个输入提示。数据集是从英语版本的IFEval数据集专业翻译而来，旨在提高加泰罗尼亚语在NLP领域的支持，并允许跨语言比较。数据集的创建过程包括数据收集、翻译和验证，确保翻译的准确性和一致性。

IFEval_ca is a prompt dataset professionally translated from English to Catalan, designed to evaluate chat or instruction fine-tuned language models. The dataset comprises 541 verifiable instructions, each instance containing one input prompt. Intended for evaluation purposes, not for training language models. Provided in JSONL format, each row contains an instance identifier and the corresponding input prompt. The translation process followed specific guidelines to ensure linguistic accuracy and relevance. Curated by the Language Technologies Unit at BSC-CNS and funded by ILENIA. The dataset is licensed under CC-BY-4.0.