lukasplevac/alpaca-cs

该数据集是流行Alpaca数据集的捷克语版本，基于yahma/alpaca-cleaned数据集创建。包含翻译成捷克语的指令、输入和输出，适合用于训练和评估LLM模型。数据格式包括instruction（指令文本，描述模型应执行的任务）、input（可选的指令输入文本）和output（模型的预期输出）。数据集大小约为50k行，语言为捷克语。翻译通过LLM（gemma3:12b）自动完成，并进行了手动质量检查。数据集适用于捷克语LLM模型的实验和指令调优。

This dataset is a Czech version of the popular Alpaca dataset, originally based on [`yahma/alpaca-cleaned`](https://huggingface.co/datasets/yahma/alpaca-cleaned). It contains **instructions, inputs, and outputs** translated into Czech, suitable for training and evaluating LLM models. The data format includes `instruction` (text describing what the model should do), `input` (optional input text for the instruction), and `output` (the expected output of the model). The dataset size is approximately 50k lines, and the language is Czech. Translations were created automatically using LLM (gemma3:12b) with manual quality checks on a sample of the data. The dataset is suitable for experiments with Czech LLM models and instruction-tuning.