stephantulkens/triviaqa-query-mxbai-pooled
收藏Hugging Face2025-10-13 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/stephantulkens/triviaqa-query-mxbai-pooled
下载链接
链接失效反馈官方服务：
资源简介：
这是基于TriviaQA数据集的查询部分，使用Mixedbread AI的mixedbread-ai/mxbai-embed-large-v1模型进行嵌入。每个文档的嵌入为1024维，并截断为每个文档的前510个token。此数据集主要适用于大规模知识蒸馏，因为它简化了像sentence-transformers这样的模型的训练过程，无需手动截断和匹配文本。
This is the query portion of the TriviaQA dataset, embedded using Mixedbread AIs mixedbread-ai/mxbai-embed-large-v1 model. Each document is embedded with 1024-dimensional embeddings, truncated to the first 510 tokens of each document. This dataset is primarily useful for large-scale knowledge distillation, as it simplifies the training process for models like sentence-transformers by avoiding manual text truncation and matching.
提供机构：
stephantulkens
