Adaptive deadline bounded query execution using generative model(s)

A method implemented by one or more processors, the method comprising:

• receiving an input query that is generated based on user interface input at a client device;
• causing a prompt to be rendered at the client device or an other client device, wherein the prompt elicits user input indicative of a deadline for generating a comprehensive response which is responsive to the input query;
• receiving, responsive to the prompt, subsequent user input;
• determining the deadline based on the received subsequent user input;
• generating an execution scheme to comply with the deadline for generating a comprehensive response which is responsive to the input query, generating the execution scheme comprising:
• decomposing the input query, decomposing the input query comprising processing the input query, using a first generative model, to determine:
• a plurality of sub-queries, and
• for each of the sub-queries, one or more corresponding tools to utilize in processing the sub-query;
• and
• generating the execution scheme based on the plurality of sub-queries, wherein each sub-query is associated with an estimated execution time, and wherein generating the execution scheme comprises generating the execution scheme such that the total of the estimated execution times for each of the sub-queries complies with the deadline,
• wherein generating the execution scheme comprises generating the execution scheme such that processing of each sub-query is to be performed based on the corresponding one or more dependencies;
• based on the execution scheme, and for each of the sub-queries:
• processing the sub-query, using the one or more corresponding tools for the sub-query, to generate one or more corresponding sub-query responses;
• in response to determining that each of the sub-queries of the execution scheme have been processed, generating the comprehensive response to the input query, generating the comprehensive response comprising processing, using the first generative model and/or a second generative model, the one or more corresponding sub-query responses for each of the sub-queries;
• and
• causing the comprehensive response to be rendered, at the client device or the other client device determined to be associated with the client device, as responsive to the input query.

A method implemented by one or more processors, the method comprising:

• receiving an input query that is generated based on user interface input at a client device;
• generating an execution scheme to comply with a deadline for generating a comprehensive response which is responsive to the input query, generating the execution scheme comprising:
• decomposing the input query, decomposing the input query comprising processing the input query, using a first generative model, to determine:
• a plurality of sub-queries, and
• for each of the sub-queries, one or more corresponding tools to utilize in processing the sub-query;
• and
• generating the execution scheme based on the plurality of sub-queries, wherein each sub-query is associated with an estimated execution time, and wherein generating the execution scheme comprises generating the execution scheme such that the total of the estimated execution times for each of the sub-queries complies with the deadline,
• wherein the execution scheme specifies one or more of:
• a level of self-evaluation to be performed to generate the comprehensive response, and
• an indication of the generative model to be used for generating the comprehensive response;
• based on the execution scheme, and for each of the sub-queries:
• processing the sub-query, using the one or more corresponding tools for the sub-query, to generate one or more corresponding sub-query responses;
• in response to determining that each of the sub-queries of the execution scheme have been processed, generating the comprehensive response to the input query, generating the comprehensive response comprising processing, using the first generative model and/or a second generative model, the one or more corresponding sub-query responses for each of the sub-queries;
• and
• causing the comprehensive response to be rendered, at the client device or another client device determined to be associated with the client device, as responsive to the input query.