atlasia/tulu-v2-sft-mixture-role-split-english-darija-cleaned
收藏Hugging Face2025-01-19 更新2025-02-15 收录
下载链接:
https://hf-mirror.com/datasets/atlasia/tulu-v2-sft-mixture-role-split-english-darija-cleaned
下载链接
链接失效反馈官方服务:
资源简介:
Tulu-v2 Role-Split English-Darija 数据集是一个经过去重和代码清理的英语-达里语角色分割对话数据集。它包含了对话中的角色(用户或助手)所使用的英语内容和对应的达里语翻译。数据集来源于多个原始数据集(例如 flan_v2、sharegpt 等),并为每个样本提供了唯一标识符。这种格式便于针对对话型语言模型和翻译模型进行角色特定的分析和训练。
The Tulu-v2 Role-Split English-Darija dataset is a deduplicated and code-free English-Darija role-separated dialogue dataset. It includes the English content and the corresponding Darija translation used by the roles in the dialogue (either user or assistant). The dataset is sourced from multiple original datasets (e.g., flan_v2, sharegpt) and provides a unique identifier for each sample. This format facilitates role-specific analysis and training for dialogue-based language models and translation models.
提供机构:
atlasia



