公开(公告)号：US20240303161A1

公开(公告)日：2024-09-12

申请号：US18592791

申请日：2024-03-01

Applicant: MOREH CORP.

Inventor： Gangwon Jo ,  Jungho Park

IPC: G06F11/14

CPC classification number: G06F11/1407 ,  G06F11/1461

Abstract: A method for performing checkpointing for fault tolerance of an application is provided, which is performed by one or more processors, and includes storing accelerator data associated with an application at a specific time point as a checkpoint, storing operation data associated with the application performed after the specific time point, and performing application fault tolerance based on the checkpoint and the stored operation data.

公开(公告)号：US20240303057A1

公开(公告)日：2024-09-12

申请号：US18427074

申请日：2024-01-30

Applicant: MOREH CORP.

Inventor： Gangwon Jo ,  Jungho Park

IPC: G06F8/41

CPC classification number: G06F8/456

Abstract: A method for pipeline parallelism compiling is provided, which is executed by one or more processors, and includes receiving a source program associated with training of a machine learning model, determining, based on the source program, a plurality of operation groups including operations executed on training data of the machine learning model, generating a plurality of micro-batches from the training data, and determining, for each of the plurality of micro-batches, a plurality of operation sets corresponding to the plurality of operation groups.

公开(公告)号：US12106076B2

公开(公告)日：2024-10-01

申请号：US17853090

申请日：2022-06-29

Applicant: MOREH CORP.

Inventor： Jaejin Lee ,  Wookeun Jung

IPC: G06F9/44 ,  G06F8/33 ,  G06N20/00

CPC classification number: G06F8/33 ,  G06N20/00

Abstract: The present disclosure relates to a method for generating a program for use in an accelerator for deep learning. The method may include receiving, by a computing device, a deep learning application, generating an element-wise operation list included in the deep learning application, generating an intermediate expression from the element-wise operation list, and generating, based on the intermediate expression, a program for use in an accelerator for the deep learning application.

公开(公告)号：US20220326922A1

公开(公告)日：2022-10-13

申请号：US17853028

申请日：2022-06-29

Applicant: MOREH CORP.

Inventor： Jaejin Lee ,  Wookeun Jung

IPC: G06F8/41 ,  G06N20/00

Abstract: The present disclosure relates to a method for automatically optimizing a program based on reinforcement learning. The method for automatically optimizing a program based on reinforcement learning includes (a) receiving an input for a source program, which includes a fixed parameter and variable parameter, (b) generating the source program based on the received input, (c) converting the source program into an object program, (d) executing the converted object program to measure a performance of the executed object program, (e) inputting the variable parameter and the measured performance into a machine learning model, and outputting a variation of the variable parameter, and (f) regenerating a source program reflecting the variation of the variable parameter.

公开(公告)号：US20240303138A1

公开(公告)日：2024-09-12

申请号：US18422655

申请日：2024-01-25

Applicant: MOREH CORP

Inventor： Gangwon Jo ,  Jungho Park

IPC: G06F9/54

CPC classification number: G06F9/542

Abstract: Provided is a method for parallelly processing a call list associated with an artificial intelligence calculation performed by one or more processors, including acquiring an original call list including a plurality of primitive operations, determining a number of accelerators to parallelly process the original call list, creating a plurality of sub call lists based on the determined number of accelerators and the original call list, and transmitting each of the created plurality of sub call lists to each of a plurality of accelerators corresponding to the determined number.

公开(公告)号：US12026487B2

公开(公告)日：2024-07-02

申请号：US17853028

申请日：2022-06-29

Applicant: MOREH CORP.

Inventor： Jaejin Lee ,  Wookeun Jung

IPC: G06F11/36 ,  G06F8/41 ,  G06N3/04 ,  G06N3/08 ,  G06N20/00 ,  G06F8/30 ,  G06F8/35 ,  G06F8/77

CPC classification number: G06F8/443 ,  G06N20/00 ,  G06F8/30 ,  G06F8/35 ,  G06F8/77

Abstract: A method for automatically optimizing a program based on reinforcement learning includes (a) receiving an input for a source program, which includes a fixed parameter and variable parameter, (b) generating the source program based on the received input, (c) converting the source program into an object program, (d) executing the converted object program to measure a performance of the executed object program, (e) inputting the variable parameter and the measured performance into a machine learning model, and outputting a variation of the variable parameter, and (f) regenerating a source program reflecting the variation of the variable parameter.

公开(公告)号：US20240118878A1

公开(公告)日：2024-04-11

申请号：US18542544

申请日：2023-12-15

Applicant: MOREH CORP. ,  Seoul National University R&DB Foundation

Inventor： Jaejin Lee ,  Jungho Park ,  Gangwon Jo ,  Heehoon Kim ,  Jinpyo Kim

IPC: G06F8/41 ,  G06N3/042

CPC classification number: G06F8/443 ,  G06N3/042

Abstract: A method for determining optimization applicability on an intermediate representation from a program is performed by one or more processors, and includes receiving, as a query, a subgraph of the intermediate representation that is a subject of determination of optimization applicability, determining a validity of the query, and if the query is valid, determining optimization applicability on the subgraph, in which the program includes data and a plurality of operations, and the intermediate representation includes a plurality of data nodes, a plurality of operation nodes, and a plurality of edges representing input/output relationships between the plurality of data nodes and the plurality of operation nodes.

公开(公告)号：US20240118876A1

公开(公告)日：2024-04-11

申请号：US18542563

申请日：2023-12-15

Applicant: MOREH CORP. ,  Seoul National University R&DB Foundation

Inventor： Jaejin Lee ,  Jungho Park ,  Gangwon Jo ,  Heehoon KIM ,  Jinpyo KIM

IPC: G06F8/41

CPC classification number: G06F8/41

Abstract: A method for managing an intermediate representation from a program is executed by one or more processors, and includes extracting, from the program, information on data for input and output and information on operation, generating an intermediate representation from the program using the extracted information on data and the extracted information on operation, storing, in a database, a corresponding relationship between the program and the intermediate representation, storing execution information on operation of the intermediate representation, and deleting at least a part of the intermediate representation based on the execution information.

公开(公告)号：US20220326915A1

公开(公告)日：2022-10-13

申请号：US17853090

申请日：2022-06-29

Applicant: MOREH CORP.

Inventor： Jaejin Lee ,  Wookeun Jung

IPC: G06F8/33 ,  G06N20/00

Abstract: The present disclosure relates to a method for generating a program for use in an accelerator for deep learning. The method may include receiving, by a computing device, a deep learning application, generating an element-wise operation list included in the deep learning application, generating an intermediate expression from the element-wise operation list, and generating, based on the intermediate expression, a program for use in an accelerator for the deep learning application.

公开(公告)号：US20240303164A1

公开(公告)日：2024-09-12

申请号：US18592839

申请日：2024-03-01

Applicant: MOREH CORP.

Inventor： Gangwon Jo ,  Jungho Park

IPC: G06F11/14

CPC classification number: G06F11/1461 ,  G06F11/1469

Abstract: A network system for running distributed environment applications is provided, including a main network part for interface between devices belonging to a plurality of device groups, and a sub-network part for interface between devices belonging to sub-device group, in which the sub-device group includes at least some of the devices belonging to each of different device groups.