Generating automated assistant responses and/or actions directly from dialog histories and resources

Abstract

The present disclosure relates to generating automated assistant responses and / or actions directly from dialog histories and resources. The invention trains and / or utilizes a single neural network model to generate a corresponding automated assistant natural language response and / or a corresponding automated assistant action for each of a plurality of assistant rounds of a dialog session between the user and the automated assistant. For example, at a given assistant round of a dialog session, corresponding natural language responses and corresponding actions can be generated jointly and directly based on output generated using a single neural network model. Corresponding responses and / or corresponding actions may be generated based on processing the dialog history and the plurality of discrete resources using a neural network model. For example, a neural network model may be used to generate responses and / or actions on a token-by-token basis.

Description

technical field [0001] The present disclosure relates to generating automated assistant responses and / or actions directly from dialog history and resources. Background technique [0002] People can use interactive software applications referred to herein as "automated assistants" (also referred to as "digital agents," "interactive personal assistants," "intelligent personal assistants," "assistant applications," "conversational agents," etc.) machine dialogue. For example, humans (who may be referred to as "users" when they interact with automated assistants) may provide commands and And / or request, in some cases can convert spoken natural language input (i.e. utterance) to text and then process it. The automated assistant responds to the request by providing a responsive user interface output and / or taking a responsive action. Responsive user interface output may include natural language output, which may optionally be converted to synthesized speech and rendered. Respo...

AI Technical Summary

Problems solved by technology

However, component-dependent pipelines can make automated assistants difficult to scale

Also, each individual component must be trained individually, which can require significant engineering effort and heavy use of computer resources during training

Furthermore, global augmentation signals (i.e., those that depend on the overall outcome of dialogue turns or complete dialogues) can be problematic and / or impossible to use effectively for training individual components

Images