US 11,960,515 B1Granted

Edge computing units for operating conversational tools at local sites

AssigneeArmada Systems Inc.

InventorsVenkata Bhanu Teja Pallakonda, Pragyana K. Mishra

Life of this patent

Estimating the term…

Loading patent…

Abstract

Computing units provided at local sites or edge locations are programmed to execute conversational tools that generate pertinent, domain-specific responses to queries received from workers at such sites or locations. The conversational tools are large language models that are trained with domain-specific knowledge documents. Data representing queries are received from workers at such sites or locations and provided as inputs to the conversational tools along with text representing nearest knowledge documents from a knowledge base associated with the domain, as well as contextual data. Responses identified based on outputs received from the conversational tools in response to the inputs are provided to the workers that generated the queries. Where subsequent queries are received from the workers, responses to the subsequent queries are identified based on the subsequent queries, nearest knowledge documents, contextual data, and conversational histories including previously received queries and responses to such queries.

Claims20

Dependent claims

An edge computing unit comprising a containerized system having:

at least one processor unit;
at least one server rack;
at least one power unit;
at least one environmental control system;
and
at least one isolation system,
wherein the at least one server rack comprises at least one data store programmed with:
one or more sets of instructions;
code for executing a first model, wherein the first model is a model configured to generate an embedding representing a semantic descriptor of a document in a vector space having a predetermined number of dimensions;
code for executing a second model, wherein the second model is a conversational model configured to generate a response to a query in reply to a prompt comprising information regarding the query;
and
a knowledge base comprising a first plurality of knowledge documents and a first plurality of knowledge embeddings, wherein each of the first plurality of knowledge documents comprises a question pertaining to a domain and an answer to the question, and wherein each of the first plurality of knowledge embeddings was generated by the first model based at least in part on one of the first plurality of knowledge documents,
wherein the one or more sets of instructions, when executed by the at least one processor unit, cause the edge computing unit to at least:
receive a first set of text representing a first query from a computer system in communication with the edge computing unit;
provide at least a portion of the first set of text as a first input to the first model;
generate a first embedding based at least in part on a first output received in response to the first input;
identify a second plurality of embeddings, wherein each of the second plurality of embeddings is one of a predetermined number of the first plurality of embeddings most similar to the first embedding;
identify a second plurality of documents, wherein each of the second plurality of documents is one of the first plurality of documents from which one of the second plurality of embeddings was generated;
generate a first prompt based at least in part on the portion of the first set of text and the second plurality of documents;
provide the first prompt as a second input to the second model;
identify a first response to the first query based at least in part on a second output received in response to the second input;
and
transmit a second set of text representing the first response to the computer system.

20 more elements

An edge computing unit comprising:

at least one processor unit;
at least one server rack;
at least one power unit;
at least one environmental control system;
and
at least one isolation system,
wherein the edge computing unit is programmed with one or more sets of instructions that, when executed by the at least one processor unit, cause the edge computing unit to at least:
receive first data representing a first query from a first system provided at a location;
generate a first embedding based at least in part on the first data;
identify a first plurality of embeddings based at least in part on the first embedding, wherein each of the first plurality of embeddings is one of a second plurality of embeddings of a knowledge base relating to a domain, and wherein each of the first plurality of embeddings is identified as similar to the first embedding according to at least one similarity metric;
generate a first prompt based at least in part on the first data and a first plurality of documents, wherein each of the first plurality of documents comprises at least a portion of one of a second plurality of documents of the knowledge base, wherein each of the second plurality of embeddings was generated based at least in part on one of the second plurality of documents, and wherein each of the second plurality of embeddings is stored in association with one of the second plurality of documents;
provide at least the first prompt as a first input to a first model;
identify a first response to the first query based at least in part on a first output received from the first model in reply to the first input;
and
transmit second data representing the first response to one of the first system or a second system provided at the location.

12 more elements

An edge computing unit provided at a local site, wherein the edge computing unit comprises a system having:

at least one processor unit;
at least one server rack;
at least one power unit;
at least one environmental control system;
and
at least one isolation system,
wherein the edge computing unit is programmed with one or more sets of instructions that, when executed by the at least one processor unit, cause the edge computing unit to at least:
receive first data representing a first query of a worker at the local site, wherein the first data is received from a computer system associated with the worker, and wherein the first data comprises at least one of:
audio data captured using the computer system, wherein the audio data represents at least one word spoken by the worker;
a set of text entered by the worker into the computer system;
and
an image captured by the worker using the computer system;
identify a first set of text associated with the first data;
generate a first embedding according to a first model based at least in part on the first set of text, wherein the first embedding is a semantic descriptor of the first set of text;
identify at least a second embedding based at least in part on the first embedding, wherein the second embedding is determined to be similar to the first embedding according to at least one similarity metric;
identify a second set of text pertaining to a domain based at least in part on the second embedding, wherein the second embedding was generated according to the first model based at least in part on the second set of text;
generate a first prompt based at least in part on the first set of text and the second set of text;
identify a first response to the first query based at least in part on the first prompt, wherein the first response comprises a third set of text;
identify second data based at least in part on the third set of text, wherein the second data comprises:
audio data identified based at least in part on the third set of text;
at least some of the third set of text;
and
an image identified based at least in part on the third set of text;
and
transmit at least a portion of the second data to the computer system.

22 more elements

Dependencies are inferred from the claim text.