bigstupidhats/dynasample_train_scoreby3llms

该数据集是一个多语言对话数据集，包含阿拉伯语、德语、英语、西班牙语、印地语、俄语、斯瓦希里语和中文等多种语言的对话内容。每个对话都有详细的元数据，如指令、输出、语言代码、用户ID、对话ID等。此外，数据集还包含了多个模型的评估指标，如Qwen2.5-7B、Llama-3.1-8B和EuroLLM-9B的损失值、困惑度等。数据集还提供了对话内容的毒性检测和审核信息，包括OpenAI和Detoxify的审核结果。

This dataset is a multilingual dialogue dataset containing conversations in multiple languages such as Arabic, German, English, Spanish, Hindi, Russian, Swahili, and Chinese. Each conversation includes detailed metadata such as instruction, output, language code, user ID, conversation ID, etc. Additionally, the dataset contains evaluation metrics for multiple models, including Qwen2.5-7B, Llama-3.1-8B, and EuroLLM-9B, with metrics like loss and perplexity. The dataset also provides toxicity detection and moderation information for the conversation content, including moderation results from OpenAI and Detoxify.