SEMANTIC REPRESENTATION MODEL PROCESSING METHOD, DEVICE, ELECTRONIC EQUIPMENT AND STORAGE MEDIUM

摘要

The present application discloses a method for processing a semantic representation model, an apparatus, an electronic device, and a storage medium, and relates to the field of artificial intelligence technology. A specific implementation method may include: collecting a training corpus including a plurality of training language materials; and performing training on the semantic representation model based on at least one of morphology, grammar and semantics using the training corpus. In the present application, by constructing unsupervised or weakly supervised pre-training tasks of three different levels of morphology, grammar, and semantics, the semantic representation model learns different classes of knowledge, such as vocabulary, grammar, and semantics, among large-scale data. By doing so, the ability of general-purpose semantic expression was strengthened, and the processing effect of NLP tasks was improved.