stephantulkens/english-word-definitions-mxbai-pooled
收藏Hugging Face2025-10-17 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/stephantulkens/english-word-definitions-mxbai-pooled
下载链接
链接失效反馈官方服务：
资源简介：
这是由Mixedbread AI的mxbai-embed-large-v1嵌入的英语单词定义数据集。每个文档使用前510个标记进行嵌入，并且由于模型的训练方法，嵌入被截断。该数据集主要用于大规模知识蒸馏，包含466k行，每行有三个键：id、嵌入和文本。文本被截断为模型实际看到的切片，这使得它可以直接用于sentence-transformers的训练，而无需手动截断文本。
This is the English Word Definitions dataset, embedded with Mixedbread AIs mxbai-embed-large-v1. Each document is embedded using the first 510 tokens, and the embeddings are truncated due to the models training method. The dataset is useful for large-scale knowledge distillation and consists of 466k rows with three keys: id, embedding, and text. The text is truncated to the slice seen by the model, making it suitable for training in sentence-transformers without manual text truncation.
提供机构：
stephantulkens
