Data management system and method suitable for artificial intelligence model and related device

Abstract

The invention discloses a data management system and method suitable for an artificial intelligence model and a related device, and the system comprises a man-machine interaction module which is used for receiving input configuration information, and the configuration information comprises a code warehouse address and hardware resource information, the code repository address comprises at least one of a first code address used for training the artificial intelligence model, a second code address used for performing offline evaluation on the trained artificial intelligence model, and a third code address used for performing simulation evaluation on the trained artificial intelligence model; the model training module is used for performing model training based on the first code address and the hardware resource information to obtain a trained artificial intelligence model; the model evaluation module is used for carrying out offline evaluation on the trained artificial intelligence model based on the second code address, and / or carrying out simulation evaluation on the trained artificial intelligence model based on the third code address; and the monitoring management module is used for monitoring and managing the operation condition of the system.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to a data management system, method and related devices suitable for artificial intelligence models. Background Art

[0002] With the rapid development of AI (Artificial Intelligence) technology, more and more application scenarios can improve data processing efficiency by using AI models and provide more traversal for humans. For example, the deep learning technology of AI models can be used to achieve autonomous driving and reduce the driving pressure of drivers.

[0003] The process from training to publishing an AI model usually includes: preparing training data, configuring the training environment, submitting training tasks, manually monitoring whether the training tasks are completed, manually exporting the trained model, and importing the manually exported model into the evaluation program for evaluation. After the evaluation passes, the model is published for application. As can be seen, the above process is cumbersome for users and consumes a lot of time and energy. Summary of the invention

[0004] The present application provides a data management system, method and related devices suitable for artificial intelligence models, which can provide a one-stop system for managing artificial intelligence models. It is not only easy to operate for users, but also can save manpower from model training to evaluation, and improve processing efficiency.

[0005] The specific technical solutions are as follows:

[0006] In a first aspect, an embodiment of the present application provides a data management system suitable for an artificial intelligence model, the system comprising:

[0007] A human-computer interaction module, configured to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing an offline evaluation on the trained artificial intelligence model, and a third code address for performing a simulation evaluation on the trained artificial intelligence model;

[0008] A model training module, configured to perform model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model when the code repository address includes the first code address;

[0009] A model evaluation module, configured to perform offline evaluation on the trained artificial intelligence model based on the second code address when the code warehouse address includes the second code address, and / or perform simulation evaluation on the trained artificial intelligence model based on the third code address when the code warehouse address includes the third code address;

[0010] The monitoring and management module is used to monitor and manage the operation of the system.

[0011] It can be seen from the above scheme that the user only needs to input configuration information in the human-computer interaction module of the system provided in the embodiment of the present application, and then the system will run the human-computer interaction module, model training module, and model evaluation module in sequence to realize the training and evaluation of the artificial intelligence model, and the monitoring management module in the system monitors and manages the operation of the system in real time without manual monitoring, thereby realizing a one-stop management system for artificial intelligence models, which is not only simple to operate for users, but also saves manpower from model training to evaluation, thereby improving processing efficiency.

[0012] In a first possible implementation manner of the first aspect, the monitoring management module is used to

[0013] When a task fails to run, a warning message of the task failure is sent to the human-computer interaction module so that the human-computer interaction module outputs the warning message; and / or,

[0014] When monitoring that a task is successfully executed, sending a reminder message of successful execution of the task to the human-computer interaction module so that the human-computer interaction module outputs the reminder message; and / or,

[0015] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0016] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0017] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0018] Through the above scheme, it can be seen that the embodiment of the present application can let users know the final running results of each task in time without manual monitoring by pushing warning information of task failure and reminder information of task success to users. It can also release the hardware resources occupied by the failed task when the task fails, and release the hardware resources occupied by the successful task when the task succeeds. When a hardware resource is idle for too long, the corresponding hardware resource is released, thereby improving resource utilization.

[0019] In a second possible implementation manner of the first aspect, the system further includes:

[0020] The log recording module is used to generate and record log information about the operation status of the system.

[0021] It can be seen from the above scheme that the embodiment of the present application can also generate and record log information about the system operation status, so that the user can query the log to know the operation status of each task, such as the reason why a task failed to run.

[0022] In a third possible implementation of the first aspect, the configuration information also includes at least one of the task submitter name, the task name, the branch name required for pulling the code, the commit ID required for pulling the code, the token required for pulling the code, the queue name where the task is located, the task execution command, and the application container engine docker image.

[0023] In a fourth possible implementation of the first aspect, the human-computer interaction module is further used to receive a query field value input for a query field on the task query interface after outputting the task query interface, and output a task query result, wherein the query field is set according to any information recorded during the execution of any task by the system.

[0024] It can be seen from the above scheme that the human-computer interaction module provided in the embodiment of the present application outputs a task query interface to the user, receives the query field value input for the query field on the task query interface, and outputs the task query result, thereby enabling the user to query or trace any task-related information, avoiding the problem of difficult information query or tracing.

[0025] In a second aspect, an embodiment of the present application provides a data management method suitable for an artificial intelligence model, the method being applied to a data management system suitable for an artificial intelligence model, the method comprising:

[0026] Receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing offline evaluation on the trained artificial intelligence model, and a third code address for performing simulation evaluation on the trained artificial intelligence model;

[0027] When the code warehouse address includes the first code address, performing model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model;

[0028] When the code warehouse address includes the second code address, an offline evaluation is performed on the trained artificial intelligence model based on the second code address, and / or, when the code warehouse address includes the third code address, a simulation evaluation is performed on the trained artificial intelligence model based on the third code address;

[0029] During the operation of the system, the operation status of the system is monitored and managed.

[0030] In a first possible implementation manner of the second aspect, the monitoring and managing the operation of the system includes:

[0031] When a task fails to run, a warning message of the task failure is generated and output; and / or,

[0032] When monitoring that a task is successfully executed, generating and outputting a reminder message that the task is successfully executed; and / or,

[0033] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0034] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0035] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0036] In a second possible implementation manner of the second aspect, the method further includes:

[0037] Generate and record log information about the operation of the system.

[0038] In a third possible implementation of the second aspect, the configuration information also includes at least one of the task submitter name, task name, branch name required for pulling code, commit ID required for pulling code, token required for pulling code, queue name where the task is located, task execution command, and application container engine docker image.

[0039] In a fourth possible implementation manner of the second aspect, the method further includes:

[0040] After outputting the task query interface, receiving the query field value input for the query field on the task query interface, and outputting the task query result, wherein the query field is set according to any information recorded during the execution of any task by the system.

[0041] It can be seen from the above scheme that the user only needs to input configuration information in the system provided in the embodiment of the present application, and then the system will implement model training and model evaluation in sequence according to the configuration information. During this period, the system can also monitor and manage the operation of the system in real time without manual monitoring, thereby realizing a one-stop management system for artificial intelligence models, which is not only simple to operate for users, but also can save manpower from model training to evaluation, and improve processing efficiency.

[0042] In a third aspect, an embodiment of the present application provides a data management device suitable for an artificial intelligence model, the device comprising:

[0043] A receiving unit, configured to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing an offline evaluation on the trained artificial intelligence model, and a third code address for performing a simulation evaluation on the trained artificial intelligence model;

[0044] a training unit, configured to perform model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model when the code repository address includes the first code address;

[0045] An evaluation unit, configured to perform an offline evaluation on the trained artificial intelligence model based on the second code address when the code warehouse address includes the second code address, and / or to perform a simulation evaluation on the trained artificial intelligence model based on the third code address when the code warehouse address includes the third code address;

[0046] The monitoring and management unit is used to monitor and manage the operation of the system during the operation of the system.

[0047] In a first possible implementation manner of the third aspect, the monitoring management unit is used to:

[0048] When a task fails to run, a warning message of the task failure is generated and output; and / or,

[0049] When monitoring that a task is successfully executed, generating and outputting a reminder message that the task is successfully executed; and / or,

[0050] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0051] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0052] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0053] In a second possible implementation manner of the third aspect, the device further includes:

[0054] A recording unit is generated, used to generate and record log information about the operation status of the system.

[0055] In a third possible implementation of the third aspect, the configuration information also includes at least one of the task submitter name, the task name, the commit ID required to pull the code, the token required to pull the code, the queue name where the task is located, the task execution command, and the application container engine docker image.

[0056] In a fourth possible implementation manner of the third aspect, the device further includes:

[0057] The receiving unit is further used to receive a query field value input for a query field on the task query interface after outputting the task query interface, wherein the query field is set according to any information recorded during the execution of any task by the system;

[0058] Output unit, used to output task query results.

[0059] It can be seen from the above scheme that the user only needs to input configuration information in the system provided in the embodiment of the present application, and then the system will implement model training and model evaluation in sequence according to the configuration information. During this period, the system can also monitor and manage the operation of the system in real time without manual monitoring, thereby realizing a one-stop management system for artificial intelligence models, which is not only simple to operate for users, but also can save manpower from model training to evaluation, and improve processing efficiency.

[0060] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any possible implementation manner of the second aspect.

[0061] In a fifth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising:

[0062] one or more processors;

[0063] The processor is coupled to a storage device, and the storage device is used to store one or more programs;

[0064] When one or more programs are executed by one or more processors, the electronic device implements the method described in any possible implementation manner of the second aspect.

[0065] In a sixth aspect, an embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed on a computer or a processor, the computer or the processor executes the method described in any possible implementation of the second aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art description are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work.

[0067] Figure 1 A schematic diagram of the structure of a data management system suitable for an artificial intelligence model provided in an embodiment of the present application;

[0068] Figure 2 An example diagram of a configuration information filling interface provided in an embodiment of the present application;

[0069] Figure 3 A flowchart of a data management method suitable for an artificial intelligence model provided in an embodiment of the present application;

[0070] Figure 4 A block diagram of a data management device suitable for an artificial intelligence model provided in an embodiment of the present application;

[0071] Figure 5 A schematic diagram of the structure of an electronic device or computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0073] It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The terms "including" and "having" in the embodiments of the present application and the accompanying drawings and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0074] Figure 1 The structure diagram of a data management system suitable for artificial intelligence models is shown in FIG. 1 , which can be applied to electronic devices or computer devices, and can be applied to terminals or servers. Specifically, the system may include:

[0075] The human-computer interaction module 110 is used to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing offline evaluation on the trained artificial intelligence model, and a third code address for performing simulation evaluation on the trained artificial intelligence model;

[0076] A model training module 120, configured to perform model training based on the first code address and hardware resource information to obtain a trained artificial intelligence model when the code repository address includes the first code address;

[0077] A model evaluation module 130, configured to perform offline evaluation on the artificial intelligence model based on the second code address when the code repository address includes the second code address, and / or to perform simulation evaluation on the artificial intelligence model based on the third code address when the code repository address includes the third code address;

[0078] The monitoring and management module 140 is used to monitor and manage the operation of the system.

[0079] In the embodiments of the present application, offline evaluation refers to simply inputting data into the artificial intelligence model to obtain model output data, but not running it in a simulated scene, let alone running it in a real scene. For example, when the artificial intelligence model is a decision-making planning model in the field of autonomous driving, offline evaluation refers to simply inputting data into the decision-making planning model to obtain model output data, but not running it in a simulated driving scene, let alone running it on a real vehicle. The indicators of offline evaluation include evaluation indicators including accuracy, recall rate, F1 value, etc. F1 value refers to a comprehensive evaluation index, F1 value = accuracy * recall rate * 2 / (accuracy + recall rate).

[0080] Simulation evaluation refers to building various scenarios for operation and running artificial intelligence models in various scenarios to test whether the output results of the artificial intelligence model can achieve the expected results. For example, when the artificial intelligence model is a decision-making planning model in the field of autonomous driving, simulation evaluation refers to building various driving scenarios and running the decision-making planning model in various driving scenarios to test whether the trajectory output by the decision-making planning model can achieve the expected trajectory. After the model evaluation module 130 completes the model evaluation, it can also generate an evaluation report and send the evaluation report to the human-computer interaction module 110 so that the human-computer interaction module 110 outputs the evaluation report to the user for viewing, and the evaluation report includes the evaluation results.

[0081] The code linked to the code repository address records the storage address of the original data required for training and evaluation, such as training sample sets and test sets. When executing the code, data can be read from the storage address. Hardware resource information includes the number of computer devices, the number of CPUs (Central Processing Units) contained in each computer device, and the number of GPUs (Graphics Processing Units) contained in each computer device.

[0082] In one embodiment, the configuration information also includes at least one of the task submitter name, task name, branch name required for pulling code, commit ID required for pulling code, token required for pulling code, queue name where the task is located, task execution command, and docker (application container engine) image.

[0083] The same task submitter can submit multiple tasks, each of which includes a unique task name. When the user does not fill in the task name, the system can automatically generate a task name according to the default rules. When the configuration information submitted by the user includes branch name and commit ID, you can select one of them to pull the code corresponding to the code repository address. For example, you can select branch name to pull the code by default. The task execution command refers to the command that triggers the execution of a task.

[0084] like Figure 2 As shown, when the user wants to train and evaluate the model, he can log in to the data management system suitable for the artificial intelligence model and enter the system homepage. The system homepage displays the fields to be read by the user and the field values ​​to be entered by the user in a key-value manner. These fields and field values ​​can be called configuration information. After the user enters the necessary field values, he can click the Submit button to submit the task to the system for automated processing. When the user makes an error in the input, he can also click the Reset button to reset it.

[0085] The embodiment of the present application provides a data management system suitable for artificial intelligence models. The user only needs to input configuration information in the human-computer interaction module of the system provided in the embodiment of the present application. The system can then sequentially run the human-computer interaction module, the model training module, and the model evaluation module to realize the training and evaluation of the artificial intelligence model. The monitoring management module in the system monitors and manages the operation of the system in real time without manual monitoring, thereby realizing a one-stop system for managing artificial intelligence models. Not only is the operation simple for users, but it can also save manpower from model training to evaluation, thereby improving processing efficiency.

[0086] In one embodiment, the monitoring management module 140 is used to

[0087] When a task fails to run, a warning message of the task failure is sent to the human-computer interaction module so that the human-computer interaction module outputs the warning message; and / or,

[0088] When monitoring that a task is successfully executed, sending a reminder message of the successful execution of the task to the human-computer interaction module so that the human-computer interaction module outputs the reminder message; and / or,

[0089] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0090] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0091] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0092] The output mode of the warning information and reminder information can be text, voice, or a combination of text and voice. In addition, in order to further enable users to obtain warning information and / or reminder information in a timely manner, the system can also be bound to user contact information such as mobile phone numbers and email addresses, and the monitoring management module 140 can notify users by sending text messages to users or sending email addresses.

[0093] The preset time threshold may be determined based on actual experience, such as 2 days.

[0094] The embodiment of the present application can push warning information of task failure and reminder information of task success to users, so that users can know the final running result of each task in time without manual monitoring. In addition, when a task fails, the hardware resources occupied by the failed task are released, and when the task succeeds, the hardware resources occupied by the successful task are released. When a hardware resource is idle for too long, the corresponding hardware resource is released, thereby improving resource utilization.

[0095] In one embodiment, the system further includes: a log recording module, configured to generate and record log information about system operation conditions.

[0096] The log recording module records the log information of each task executed by the system, including the start time and end time of the task. In the task, the start time, end time, whether the training is successful, and the reason for training failure of the model training performed by the model training module 120; the start time, end time, whether the evaluation is successful, and the reason for evaluation failure of the model evaluation performed by the model evaluation module 130; and the resource scheduling status performed by the monitoring management module 140, whether and when early warning information is generated, whether and when reminder information is generated, etc.

[0097] It can be seen from the above scheme that the embodiment of the present application can also generate and record log information about the system operation status, so that the user can query the log to know the operation status of each task, such as the reason why a task failed to run.

[0098] In one implementation, the human-computer interaction module 110 is further configured to receive a query field value input for a query field on the task query interface after outputting the task query interface, and output a task query result.

[0099] From the time when the human-computer interaction module 110 of the system starts to receive the configuration information input by the user, the system will record various information of each task, such as which code address each trained artificial intelligence model was trained by which user. In this case, the system can provide a query function to the user to facilitate the user to query or trace relevant information. Among them, the query field is set according to any information recorded during the execution of any task by the system. For example, the query field includes: submitter name, task name, task execution command, etc. The task query result includes at least one piece of information related to the task, including the name of the task submitter, task name, task execution command, task start time, task end time, whether the task is successful, hardware resource information occupied by the task, etc.

[0100] The human-computer interaction module provided in the embodiment of the present application outputs a task query interface to the user, receives the query field value input for the query field on the task query interface, and outputs the task query result, thereby enabling the user to query or trace any task-related information, avoiding the problem of difficult information query or tracing. For example, when a certain version of an artificial intelligence model has a problem in the application, the system can be used to query or trace the model based on the configuration information provided by the user, whether there is a problem with the training sample set, etc., and finally determine the cause of the problem.

[0101] Corresponding to the above embodiment, another embodiment of the present application provides a data management method suitable for an artificial intelligence model, the method is applied to a data management system suitable for an artificial intelligence model, such as Figure 3 As shown, the method includes:

[0102] S210: Receive input configuration information.

[0103] Among them, the configuration information includes a code warehouse address and hardware resource information. The code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for offline evaluation of the trained artificial intelligence model, and a third code address for simulation evaluation of the trained artificial intelligence model.

[0104] The configuration information also includes at least one of the task submitter name, task name, branch name required for pulling code, commit ID required for pulling code, token required for pulling code, queue name where the task is located, task execution command, and docker image.

[0105] The hardware resource information includes the number of computer devices, the number of CPUs included in each computer device, and the number of GPUs included in each computer device.

[0106] S220: When the code repository address includes the first code address, model training is performed based on the first code address and hardware resource information to obtain a trained artificial intelligence model.

[0107] The code in the first code address link includes not only the code for model training but also the storage address of the original data required for the training sample set.

[0108] S230: When the code repository address includes a second code address, an offline evaluation is performed on the trained artificial intelligence model based on the second code address, and / or, when the code repository address includes a third code address, a simulation evaluation is performed on the trained artificial intelligence model based on the third code address.

[0109] The code linked to the second code address includes not only the code for offline evaluation but also the storage address of the truth value set corresponding to the training sample set. Similarly, the code linked to the third code address includes not only the code for simulation evaluation but also the storage address of the truth value set corresponding to the training sample set.

[0110] S240: During the system operation, monitor and manage the operation status of the system.

[0111] During the system operation, you can monitor whether the task is running successfully, the hardware resource usage, etc., and issue warnings / reminders for the monitored task operation status, schedule resources, etc.

[0112] When a task fails to run, a warning message of the task failure is generated and output; and / or,

[0113] When monitoring that a task is successfully executed, generating and outputting a reminder message that the task is successfully executed; and / or,

[0114] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0115] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0116] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0117] The embodiment of the present application provides a data management method suitable for artificial intelligence models. The user only needs to input configuration information in the system provided in the embodiment of the present application, and then the system will implement model training and model evaluation in sequence according to the configuration information. During this period, the system can also monitor and manage the operation of the system in real time without manual monitoring, thereby realizing a one-stop management system for artificial intelligence models. It is not only simple to operate for users, but also can save manpower from model training to evaluation, thereby improving processing efficiency.

[0118] In one embodiment, in order to facilitate the user to know the operation status of each task, such as the reason why a task failed to run, the system can also generate and record log information about the system operation status. The log recording module records the log information of each task executed by the system, including the start time and end time of the task, the start time and end time of the model training, whether the training is successful, and the reason for the training failure, the start time and end time of the model evaluation, whether the evaluation is successful, and the reason for the evaluation failure, as well as the resource scheduling status, whether and when to generate warning information, whether and when to generate reminder information, etc.

[0119] In one embodiment, in order to avoid the problem of difficult information query or tracing, the system provided in the embodiment of the present application can also output a task query interface, receive the query field value input for the query field on the task query interface, and output the task query result, wherein the query field is set according to any information recorded during the execution of any task by the system. For example, the query field includes: submitter name, task name, task execution command, etc. The task query result includes at least one piece of information related to the task, including the task submitter name, task name, task execution command, task start time, task end time, whether the task is successful, hardware resource information occupied by the task, etc.

[0120] Based on the above embodiments, another embodiment of the present application provides a data management device suitable for an artificial intelligence model, such as Figure 4 As shown, the device comprises:

[0121] A receiving unit 310 is used to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing an offline evaluation on the trained artificial intelligence model, and a third code address for performing a simulation evaluation on the trained artificial intelligence model;

[0122] A training unit 320, configured to perform model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model when the code repository address includes the first code address;

[0123] An evaluation unit 330 is configured to perform an offline evaluation on the trained artificial intelligence model based on the second code address when the code warehouse address includes the second code address, and / or to perform a simulation evaluation on the trained artificial intelligence model based on the third code address when the code warehouse address includes the third code address;

[0124] The monitoring management unit 340 is used to monitor and manage the operation of the system during the operation of the system.

[0125] In a possible implementation, the monitoring management unit 340 is used to:

[0126] When a task fails to run, a warning message of the task failure is generated and output; and / or,

[0127] When monitoring that a task is successfully executed, generating and outputting a reminder message that the task is successfully executed; and / or,

[0128] When a task fails to run, the hardware resources occupied by the failed task are released; and / or,

[0129] When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or,

[0130] When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

[0131] In a possible implementation manner, the device further includes:

[0132] A recording unit is generated, used to generate and record log information about the operation status of the system.

[0133] In a possible implementation, the configuration information also includes at least one of the task submitter name, task name, commit ID required to pull code, token required to pull code, queue name where the task is located, task execution command, and application container engine docker image.

[0134] In a possible implementation manner, the device further includes:

[0135] The receiving unit is further used to receive a query field value input for a query field on the task query interface after outputting the task query interface, wherein the query field is set according to any information recorded during the execution of any task by the system;

[0136] Output unit, used to output task query results.

[0137] The embodiment of the present application provides a data management device suitable for artificial intelligence models. The user only needs to input configuration information in the system provided in the embodiment of the present application, and then the system will implement model training and model evaluation in sequence according to the configuration information. During this period, the system can also monitor and manage the operation of the system in real time without manual monitoring, thereby realizing a one-stop management system for artificial intelligence models. It is not only simple to operate for users, but also can save manpower from model training to evaluation, thereby improving processing efficiency.

[0138] Based on the above embodiments, another embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in any of the above embodiments is implemented.

[0139] Based on the above embodiments, another embodiment of the present application provides an electronic device or a computer device, such as Figure 5 As shown, including:

[0140] one or more processors 410;

[0141] The processor 410 is coupled to a storage device 420, and the storage device 420 is used to store one or more programs;

[0142] When the one or more programs are executed by the one or more processors 410, the electronic device or computer device implements the method described in any of the above embodiments.

[0143] Based on the above embodiments, another embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed on a computer or a processor, the computer or the processor executes the method described in any of the above embodiments.

[0144] The above device embodiment corresponds to the method embodiment and has the same technical effect as the method embodiment. For specific description, please refer to the method embodiment. The device embodiment is obtained based on the method embodiment. For specific description, please refer to the method embodiment part, which will not be repeated here. It can be understood by those skilled in the art that the accompanying drawings are only schematic diagrams of one embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application.

[0145] Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description of the embodiment, or can be changed accordingly and located in one or more devices different from the embodiment. The modules in the above embodiment can be combined into one module, or can be further divided into multiple sub-modules.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A data management system suitable for an artificial intelligence model, characterized in that: The system comprises: A human-computer interaction module, configured to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing an offline evaluation on the trained artificial intelligence model, and a third code address for performing a simulation evaluation on the trained artificial intelligence model; A model training module, configured to perform model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model when the code repository address includes the first code address; A model evaluation module, configured to perform offline evaluation on the trained artificial intelligence model based on the second code address when the code warehouse address includes the second code address, and / or perform simulation evaluation on the trained artificial intelligence model based on the third code address when the code warehouse address includes the third code address; The monitoring and management module is used to monitor and manage the operation of the system.

2. The system according to claim 1, characterized in that The monitoring management module is used to When a task fails to run, a warning message of the task failure is sent to the human-computer interaction module so that the human-computer interaction module outputs the warning message; and / or, When monitoring that a task is successfully executed, sending a reminder message of the successful execution of the task to the human-computer interaction module so that the human-computer interaction module outputs the reminder message; and / or, When a task fails to run, the hardware resources occupied by the failed task are released; and / or, When a task is monitored to be successful, the hardware resources occupied by the successful task are released; and / or, When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

3. The system according to claim 1, characterized in that The system further comprises: The log recording module is used to generate and record log information about the operation status of the system.

4. The system according to claim 1, characterized in that The configuration information also includes at least one of the task submitter name, task name, branch name required for pulling code, commit ID required for pulling code, token required for pulling code, queue name where the task is located, task execution command, and application container engine docker image.

5. The system according to any one of claims 1 to 4, characterized in that: The human-computer interaction module is also used to receive a query field value input for a query field on the task query interface after outputting the task query interface, and output a task query result, wherein the query field is set according to any information recorded during the execution of any task by the system.

6. A data management method suitable for an artificial intelligence model, characterized in that: The method is applied to a data management system suitable for an artificial intelligence model, and the method comprises: Receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing offline evaluation on the trained artificial intelligence model, and a third code address for performing simulation evaluation on the trained artificial intelligence model; When the code warehouse address includes the first code address, performing model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model; When the code warehouse address includes the second code address, an offline evaluation is performed on the trained artificial intelligence model based on the second code address, and / or, when the code warehouse address includes the third code address, a simulation evaluation is performed on the trained artificial intelligence model based on the third code address; During the operation of the system, the operation status of the system is monitored and managed.

7. The method according to claim 6, characterized in that The monitoring and management of the operation of the system includes: When a task fails to run, a warning message of the task failure is generated and output; and / or, When monitoring that a task is successfully executed, generating and outputting a reminder message that the task is successfully executed; and / or, When a task fails to run, the hardware resources occupied by the failed task are released; and / or, When a task is monitored to be successfully executed, the hardware resources occupied by the successfully executed task are released; and / or, When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

8. The method according to claim 6, characterized in that The method further comprises: Generate and record log information about the operation of the system.

9. The method according to claim 6, characterized in that The configuration information also includes at least one of the task submitter name, task name, branch name required for pulling code, commit ID required for pulling code, token required for pulling code, queue name where the task is located, task execution command, and application container engine docker image.

10. The method according to any one of claims 6 to 9, characterized in that: The method further comprises: After outputting the task query interface, receiving the query field value input for the query field on the task query interface, and outputting the task query result, wherein the query field is set according to any information recorded during the execution of any task by the system.

11. A data management device suitable for an artificial intelligence model, characterized in that: The device comprises: A receiving unit, configured to receive input configuration information, wherein the configuration information includes a code warehouse address and hardware resource information, and the code warehouse address includes at least one of a first code address for training an artificial intelligence model, a second code address for performing an offline evaluation on the trained artificial intelligence model, and a third code address for performing a simulation evaluation on the trained artificial intelligence model; a training unit, configured to perform model training based on the first code address and the hardware resource information to obtain the trained artificial intelligence model when the code repository address includes the first code address; An evaluation unit, configured to perform an offline evaluation on the trained artificial intelligence model based on the second code address when the code warehouse address includes the second code address, and / or to perform a simulation evaluation on the trained artificial intelligence model based on the third code address when the code warehouse address includes the third code address; The monitoring and management unit is used to monitor and manage the operation of the system during the operation of the system.

12. The device according to claim 11, characterized in that The monitoring management unit is used to: When a task fails to run, a warning message of the task failure is generated and output; and / or, When monitoring that a task is running successfully, generate and output reminder information that the task is running successfully; and / or, When a task fails to run, the hardware resources occupied by the failed task are released; and / or, When a task is monitored to be successful, the hardware resources occupied by the successful task are released; and / or, When it is monitored that the idle time of the hardware resources applied for a task is greater than or equal to the preset time threshold, the corresponding hardware resources are released.

13. The device according to claim 11, characterized in that The device also includes: A recording unit is generated, used to generate and record log information about the operation status of the system.

14. The device according to claim 11, characterized in that The configuration information also includes at least one of the task submitter name, task name, commit ID required to pull the code, token required to pull the code, name of the queue where the task is located, task execution command, and application container engine docker image.

15. The device according to any one of claims 11 to 14, characterized in that The device also includes: The receiving unit is further used to receive a query field value input for a query field on the task query interface after outputting the task query interface, wherein the query field is set according to any information recorded during the execution of any task by the system; Output unit, used to output task query results.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 6 to 10 is implemented.

17. An electronic device, characterized in that: The electronic device comprises: one or more processors; The processor is coupled to a storage device, and the storage device is used to store one or more programs; When the one or more programs are executed by the one or more processors, the electronic device implements the method according to any one of claims 6 to 10.

Images