# US 2021/0334078 A1 · Machine learning model representation and execution

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0334078 A1
- Document ID: US20210334078A1
- Status: Published application
- Kind: A1
- Application number: 16/858,225
- Filed: 2020-04-24
- Published: 2021-10-28
- Earliest claimed: 2020-04-24
- Assignees: AT&T Intellectual Property I, L.P.
- Applicants: AT&T Intellectual Property I, L.P.
- Inventors: Jeffrey B. Saxon; Jeffrey Dix
- Web page: https://patentgrep.com/patent/US20210334078A1

## CPC

- G06F 8/35: model driven
- G06F 8/63: {Image based installation; Cloning; Build to order}
- G06N 20/00: Machine learning
- H04L 67/34: {involving the movement of software or configuration parameters (network booting or remote initial program loading [RIPL] G06F9/4416)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Aspects of the subject disclosure may include, for example, a device, including a processing system including a processor; and a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations including receiving user specified metadata for execution tasks associated with a machine learning (ML) model; receiving artifacts specifying program code for implementing the ML model; creating a file system structure for a container to hold the ML model; receiving environment variables for operation of the ML model; and building the container including a model image for the ML model. Other embodiments are disclosed.

## Claims

### Claim 1

A device, comprising:
a processing system including a processor; and
a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations, the operations comprising:
receiving user specified metadata for execution tasks associated with a machine learning (ML) model;
receiving artifacts specifying program code for implementing the ML model;
creating a file system structure for a container to hold the ML model;
receiving environment variables for operation of the ML model; and
building the container including a model image for the ML model.

### Claim 2 (depends on 1)

The device of claim 1, wherein the user specified metadata comprises a shell command for training the ML model.

### Claim 3 (depends on 1)

The device of claim 1, wherein the user specified metadata comprises a shell command for executing the ML model.

### Claim 4 (depends on 1)

The device of claim 1, wherein the user specified metadata comprises instructions to a package manager.

### Claim 5 (depends on 1)

The device of claim 1, wherein the user specified metadata comprises hardware capabilities for executing the ML model.

### Claim 6 (depends on 5)

The device of claim 5, wherein the hardware capabilities for executing the ML model comprise a number and type of central processing units, a number and type of graphics processing units, an amount of memory, an amount of volatile memory, an amount of storage, or a combination thereof.

### Claim 7 (depends on 1)

The device of claim 1, wherein the artifacts include binary library files needed to implement the ML model.

### Claim 8 (depends on 1)

The device of claim 1, wherein the artifacts include an enumeration of values required by the ML model.

### Claim 9 (depends on 1)

The device of claim 1, wherein the file system structure comprises a path to a model artifacts folder, a data volume, and intermediate volume, a predictions volume, or a combination thereof.

### Claim 10 (depends on 1)

The device of claim 1, wherein the environment variables include a first path to a first file containing data for training the ML model, a second path to a second file containing data for testing the ML model, a third path to an intermediate volume to provide non-volatile storage for the ML model, a fourth path to a third file comprising output results, or a combination thereof.

### Claim 11 (depends on 1)

The device of claim 1, wherein the container has network connectivity during the building of the model image.

### Claim 12 (depends on 1)

The device of claim 1, wherein the building comprises layering the ML model onto a base image.

### Claim 13 (depends on 12)

The device of claim 12, wherein the building further comprises copying the artifacts into the container.

### Claim 14 (depends on 13)

The device of claim 13, wherein the building further comprises installing dependencies of the ML model into the container using a package manager.

### Claim 15 (depends on 13)

The device of claim 13, wherein the processing system comprises a plurality of processors operating in a distributed computing environment.

### Claim 16

A machine-readable medium, comprising executable instructions that, when executed by a processing system including a processor, facilitate performance of operations, the operations comprising:
creating a file system structure for a container to hold a machine learning (ML) model;
receiving environment variables for operation of the ML model;
layering a model image for the ML model on a base image in the container;
copying artifacts into the container; and
installing dependencies of the ML model into the container using a package manager.

### Claim 17 (depends on 16)

The machine-readable medium of claim 16, wherein the artifacts comprise program code for the ML model, binary files needed to execute the program code for the ML model, or a combination thereof.

### Claim 18 (depends on 17)

The machine-readable medium of claim 17, wherein the processing system comprises a plurality of processors operating in a distributed computing environment.

### Claim 19

A method, comprising:
layering, by a processing system including a processor, a base image over a generic container;
adding, by the processing system, user-specified packages wherein the user-specified packages are installed by a package manager; and
layering, by the processing system, program code for a machine learning (ML) model.

### Claim 20 (depends on 19)

The method of claim 19, wherein the base image comprises the package manager, hardware specifications, a programming language, ML toolkits, a repository manager, or a combination thereof.
