Language model parameter determination method, device and computer device

Abstract

The invention relates to a method for determining parameters of a language model. The method comprises the following steps: acquiring a plurality of sample texts; combining the plurality of sample texts and a preset thesaurus to train the original language model to obtain a training language model; training the original language model to obtain a training language model; training the training language model to obtain a training language model. Obtaining a starting word from the preset thesaurus, obtaining a generated text combining the obtained starting word and the training language model, and repeatedly obtaining a plurality of generated texts; Inputting a first preset amount of the generated text and a second preset amount of the sample text into a preset classifier to obtain a currentclassification accuracy rate; Acquiring parameters of the training language model when the current classification accuracy is a preset ratio. The method of the invention can effectively improve the optimization efficiency of the language generation model and reduce the cost.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method, device and computer equipment for determining language model parameters. Background technique [0002] Many industries in the market need a large number of text descriptions such as market articles, promotional information, shopping guide articles, and new product launches. Therefore, they are trying to use neural network language models for language writing, but there are many parameters in the neural network. Big influence, some parameters affect the accuracy rate, some parameters affect the training cost, and some parameters affect the training speed. [0003] Therefore, it is necessary to optimize the parameters of the neural network. At present, in the market, manual optimization is performed by repeatedly adjusting parameters based on past experience, but manual optimization is inefficient. Contents of the invention [0004] The purpose of the present...

AI Technical Summary

Problems solved by technology

[0002] Many industries in the market need a large number of text descriptions such as market articles, promotional information, shopping guide articles, and new product launches. Therefore, they are trying to use neural network language models for language writing, but there are many parameters in the neural network. Big impact, some parameters affect the accuracy, some parameters affect the training cost, and some parameters affect the training speed

[0003] Therefore, it is necessary to optimize the parameters of the neural network. At present, the market uses manual adjustments to optimize the parameters based on past experience, but the efficiency of manual optimization is low.

Images