Interaction model training method and device, computer equipment and storage medium

Abstract

The invention relates to an interaction model training method and device, computer equipment and a storage medium, and relates to artificial intelligence. The interaction model training method comprises the following steps: obtaining a first interaction state feature corresponding to a virtual interaction environment, and obtaining a first interaction action, the first interaction action being determined by inputting the first interaction state feature into a to-be-trained first interaction model; obtaining an income obtained by executing the first interaction action by a target virtual object, and taking the income as a first income; inputting the first interaction state feature and the first interaction action into a target strategy discrimination model corresponding to a target interaction strategy to obtain a first strategy discrimination value; performing calculation to obtain a second income according to the first strategy discrimination value; performing calculation according tothe first income and the second income to obtain a target income; and adjusting model parameters of the to-be-trained first interaction model according to the target income to obtain an updated firstinteraction model. The method can improve the model training effect.

Description

technical field

[0001] The present application relates to the technical field of artificial intelligence, in particular to an interaction model training method, device, computer equipment and storage medium. Background technique

[0002] With the continuous development of Internet technology, games have gradually become popular entertainment programs. For example, users can compete with other game players through Multiplayer Online Battle Arena (MOBA).

[0003] At present, artificial intelligence models can be used to play games. For example, when game players are offline, they can be temporarily hosted, and artificial intelligence models can be used to replace offline real players to compete with other game real players. Artificial intelligence models need to be trained using training data in advance. Most of the current artificial intelligence model training relies on continuous battle training to evolve. However, there are often problems that the trained models cannot mee...

Images