首页> 外国专利> Model improvement methods and equipment based on pre-training semantic models

Model improvement methods and equipment based on pre-training semantic models

机译:基于预培训语义模型的模型改进方法和设备

摘要

PROBLEM TO BE SOLVED: To provide a model improvement method and an apparatus based on a semantic model of pre-training in natural language processing and deep learning. The method obtains an initial improvement model based on a pre-trained semantic model, determines semantic result information of an input vector based on a hash search method, and trains the initial improvement model based on a model distillation method. , Get an improved model. The method obtains the semantic result information of the input vector by performing a hash search on the input vector, replaces the complicated iterative calculation process of the original semantic model, and creates an improved model with few model parameters and high compression ratio. It can be acquired and the processing speed of the improved model can be increased. [Selection diagram] Fig. 2
机译:要解决的问题:提供一种基于自然语言处理和深度学习预训练语义模型的模型改进方法和装置。 该方法基于预先训练的语义模型获得初始改进模型,基于哈希搜索方法确定输入向量的语义结果信息,并基于模型蒸馏方法列举初始改进模型。 ,得到一个改进的模型。 该方法通过在输入向量上执行哈希搜索来获得输入矢量的语义结果信息,取代原始语义模型的复杂迭代计算过程,并利用少量模型参数和高压缩比创建改进的模型。 可以获得它,并且可以增加改进模型的处理速度。 [选择图]图2

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号