niyatibafna/imperfect_english_prompts

这是一个用于评估用户提示中的错误和扰动如何影响大型语言模型（LLM）在机器翻译及其评估任务上性能的数据集。数据集包含了源文本、参考翻译、模型输出、源语言和目标语言等信息，并对提示应用了不同类型的噪声。数据集分为六个部分，每个部分使用不同的模型生成输出。

This dataset is for evaluating how errors and perturbations in user prompts affect the performance of large language models (LLMs) on machine translation and machine translation evaluation tasks. It includes source text, reference translation, model output, source language, and target language, with different types of noise applied to the prompts. The dataset is split into six parts, each using different models to generate outputs.