Model object management and storage system

Patent metadata

Field	Value
Publication number	US-12174800-B2
Application number	US-202418415253-A
Country	US
Kind code	B2
Filing date	Jan 17, 2024
Priority date	Mar 29, 2017
Publication date	Dec 24, 2024
Grant date	Dec 24, 2024

Abstract

A model management system provides a centralized repository for storing and accessing models. The model management system receives an input to store a model object in a first model state generated based on a first set of known variables. The model management system generates a first file including a first set of functions defining the first model state and associates the first file with a model key identifying the model object. The model management system receives an input to store the model object in a second model state having been generated based on the first model state and a second set of known variables. The model management system generates a second file including a second set of functions defining the second model state and associates the second file with the model key. The model management system identifies available versions of the model object based on the model key.

First claim

What is claimed is:

1. A model management system comprising:
one or more memories having instructions stored thereon; and
one or more processors configured to execute the instructions and perform operations comprising:
identifying, based on an input, a first model state and a second model state corresponding to a model object, the first model state being generated based on a first set of known variables, and the second model state being generated based on the model object in the first model state and a second set of known variables; and
generating a representation of data indicative of a set of model states of the model object, the set of model states including the first model state and the second model state.

2. The model management system of claim 1 , wherein the first model state of the model object includes a first set of functions, and the second model state of the model object includes a second set of functions.

3. The model management system of claim 2 , wherein the operations further comprising:
generating the first set of functions based on the first set of known variables, the first set of functions including a first parameter; and
generating the second set of functions based on the first model state and the second set of known variables, the second set of functions including an updated first parameter that is different than the first parameter.

4. The model management system of claim 3 , wherein the operations further comprise:
receiving an input to evaluate performance of the model object; and
generating a report based on the first set of functions and the second set of functions, the report indicating a change between the first parameter and the updated first parameter.

5. The model management system of claim 3 , wherein the operations further comprise:
generating a third set of functions based on the second model state and a third set of known variables, the third set of functions defining a third model state of the model object and including a second updated first parameter that is different than the updated first parameter.

6. The model management system of claim 5 , wherein the operations further comprise:
receiving an input to store the model object in the third model state;
generating a third file including the third set of functions defining the third model state; and
associating the third file with a model key identifying the model object.

7. The model management system of claim 1 , wherein the operations further comprise:
receiving an input to store the model object in the first model state;
generating a first file including a first set of functions defining the first model state;
associating the first file with a model key identifying the model object;
receiving an input to store the model object in the second model state;
generating a second file including a second set of functions defining the second model state; and
associating the second file with the model key identifying the model object.

8. The model management system of claim 7 , wherein the first file and the second file are both JavaScript Object Notation files.

9. The model management system of claim 1 , wherein the model object includes a machine learning model.

10. A method comprising:
identifying, based on an input, a first model state and a second model state corresponding to a model object, the first model state being generated based on a first set of known variables, and the second model state being generated based on the model object in the first model state and a second set of known variables; and
generating a representation of data indicative of a set of model states of the model object, the set of model states including the first model state and the second model state;
wherein the method is performed using one or more processors.

11. The method of claim 10 , wherein the first model state of the model object includes a first set of functions, and the second model state of the model object includes a second set of functions.

12. The method of claim 11 , further comprising:
generating the first set of functions based on the first set of known variables, the first set of functions including a first parameter; and
generating the second set of functions based on the first model state and the second set of known variables, the second set of functions including an updated first parameter that is different than the first parameter.

13. The method of claim 12 , further comprising:
receiving an input to evaluate performance of the model object; and
generating a report based on the first set of functions and the second set of functions, the report indicating a change between the first parameter and the updated first parameter.

14. The method of claim 12 , further comprising:
generating a third set of functions based on the second model state and a third set of known variables, the third set of functions defining a third model state of the model object and including a second updated first parameter that is different than the updated first parameter.

15. The method of claim 14 , further comprising:
receiving an input to store the model object in the third model state;
generating a third file including the third set of functions defining the third model state; and
associating the third file with a model key identifying the model object.

16. The method of claim 10 , further comprising:
receiving an input to store the model object in the first model state;
generating a first file including a first set of functions defining the first model state;
associating the first file with a model key identifying the model object;
receiving an input to store the model object in the second model state;
generating a second file including a second set of functions defining the second model state; and
associating the second file with the model key identifying the model object.

17. The method of claim 16 , wherein the first file and the second file are both JavaScript Object Notation files.

18. The method of claim 10 , wherein the model object includes a machine learning model.

19. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
identifying, based on an input, a first model state and a second model state corresponding to a model object, the first model state being generated based on a first set of known variables, and the second model state being generated based on the model object in the first model state and a second set of known variables; and
generating a representation of data indicative of a set of model states of the model object, the set of model states including the first model state and the second model state.

20. The non-transitory computer-readable medium of claim 19 , wherein the first model state of the model object includes a first set of functions, and the second model state of the model object includes a second set of functions.

Assignees

• Palantir Technologies Inc

Inventors

• Lisuk David
• Erenrich Daniel
• Xu Guodong
• Voloch Luis
• Agarwal Rahul
• Slowik Simon
• Zamoshchin Aleksandr
• Menck Andre Frederico Cavalheiro
• Mukherjee Anirvan
• Chin Daniel

Classifications

• G06F2111/20

  Configuration CAD, e.g. designing by assembling or positioning modules selected from libraries of predesigned modules · CPC title

• G06F30/00

  Computer-aided design [CAD] · CPC title

• G06F16/196

  Specific adaptations of the file system to access devices and non-file objects via standard file system access operations, e.g. pseudo file systems (dedicated interfaces to storage systems G06F3/0601) · CPC title

• G06F16/13

  File access structures, e.g. distributed indices (arrangements of input from, or output to, record carriers G06F3/06) · CPC title

• G06N7/00

  Computing arrangements based on specific mathematical models · CPC title