Microsoft Technology Licensing, LLC

Bradley Moore Abrams

Anders Hejlsberg

Patrick Henri Dussud

Christopher Wellington Brumme

Vance Palmer Morrison

James S Miller

Krzysztof J Cwalina

Brian D Harry

Scott Michael Wiltamuth

Peter F Sollich

George Harold Bosworth

Michael C Fanning

Michael Pizzo

Christopher L Anderson

Anthony J Moore

Elbio Renato Torres Abib

Robert Allan Brigham, Ii

Payal Bajaj

Kamal Ginotra

Justin Daniel Harris

Owais Khan Mohammed

Mahmoud Adada

Michael Lawrence Valenzuela

Barun Patra

Nathan Keith Walker

Vishrav Chaudhary

Hui Liu

Katherine Irene Cook

Ashutosh Devendrakumar Adhikari

Systems and methods for resource conservation based on query complexity are disclosed. An input query is received (e.g., via a chat interface) and provided to a response classifier, which is a machine-learning classifier that is trained to classify input queries with complexity scores that indicate how difficult it is likely to be for an artificial intelligence (AI) model to generate a response to the input query. If the complexity score exceeds a threshold score, the input query is provided to a first AI model (e.g., a relatively high-complexity AI model having a large number of parameters, relatively long response latencies, and/or other performance characteristics). If the complexity score does not exceed the threshold score, the input query is provided to a second AI model (e.g., a lower-complexity AI model having fewer parameters, shorter response latencies, and/or other differences in performance characteristics relative to the first AI model).

Resource conservation based on query complexity

The disclosure describes utilizing an image query system to improve response accuracy and reduce computational steps resources in responding to natural language queries of input images. In various implementations, the image query system utilizes grounding information from one or more sources to determine accurate information for an input image. For example, the image query system uses a single comprehensive image prompt to obtain extensive visual image grounding information for the input image from a visual-based large generative model. Additionally, or in alternative implementations, the image query system obtains reverse image search grounding information for the input image. The image query system then cleverly utilizes the grounding information with a large generative language model to generate text query responses to image-based queries of the input image more accurately and efficiently.

Using large generative models with improved grounding to improve image context queries

A computing system is disclosed that includes a processor and memory. The memory stores instructions that, when executed by the processor, cause the processor to perform several acts. The acts include generating a prompt that is to be input to a generative language model. The prompt includes conversational input set forth by a user. The acts further comprise providing the prompt as input to the generative language model, and receiving conversational output from the generative language model, where the generative language model generated the conversational output based upon the prompt. Additionally, the acts comprise streaming the conversational output on one of a SERP or webpage to which the user has navigated from the SERP.

Streaming of chat in SERP

A system and methods for versioning software components in connection with a computer programming language are provided. In exemplary aspects, the versioning system makes use of intelligent defaults, provides a vehicle for unambiguous specification of developer intent and implements conflict resolution rules in connection with the versioning of software components. In another aspect, the versioning system bounds names at run-time, but does not bound offsets at compile-time.

System and methods for providing versioning of software components in a computer programming language

A method and system for compiling multiple source language files that share a common library. The common library is represented in a common language that can be used by multiple different source languages. Font end compiler systems read the common language files that make up the common library and the source language files that use the library. Additionally, the front end systems produce common language files. The common language files produced by the front end systems can be used in the common library. The common language files may also be supplied to a back end system or runtime environment that further compiles the common language file to an executable form and executes the file. At runtime, the common language file is used by the runtime environment to layout the objects and methods used during execution.

Method and system for compiling multiple languages

A first exemplary method implementation for designing an application programming interface (API) includes: preparing multiple code samples for a core scenario, each respective code sample of the multiple code samples corresponding to a respective programming language of multiple programming languages; and deriving the API from the core scenario responsive to the multiple code samples. A second exemplary method for designing an API includes: selecting a core scenario for a feature area; writing at least one code sample for the core scenario; and deriving an API for the core scenario responsive to the at least one code sample. A third exemplary method for designing an API includes: deriving an API for a scenario responsive to at least one code sample written with regard to the scenario; performing one or more usability studies on the API utilizing multiple developers; and revising the API based on the one or more usability studies.

Design of application programming interfaces (APIs)

Checking program elements and program element behaviors for adherence to software development rules. An analysis engine receives a rule assembly containing metadata and intermediate language instructions representing one or more rules. The analysis engine also receives a target assembly containing metadata and intermediate language instructions representing one or more program elements and/or one or more program element behaviors. The analysis engine identifies (e.g., through reflection or retrieval of intermediate language instructions) a program element or program element behavior from the target assembly that is to be checked for adherence to a rule from the rule assembly. The analysis engine applies the rule to the program element or program element behavior to check the program element or program element behavior for adherence to the rule. Results can be stored for later viewing by a programmer.

Non-invasive rule-based binary analysis of software assemblies

Growing community of inventors

Kirkland, WA, United States of America

Bradley Moore Abrams