CN114938332A - Model parameter configuration method and device, electronic equipment and readable storage medium - Google Patents
Model parameter configuration method and device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN114938332A CN114938332A CN202210759448.6A CN202210759448A CN114938332A CN 114938332 A CN114938332 A CN 114938332A CN 202210759448 A CN202210759448 A CN 202210759448A CN 114938332 A CN114938332 A CN 114938332A
- Authority
- CN
- China
- Prior art keywords
- target preset
- model
- configuration
- feature
- features
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 239000013598 vector Substances 0.000 claims description 46
- 238000011176 pooling Methods 0.000 claims description 36
- 238000012545 processing Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 6
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000012549 training Methods 0.000 description 7
- 230000014509 gene expression Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 101100501281 Caenorhabditis elegans emb-1 gene Proteins 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Stored Programmes (AREA)
Abstract
The invention relates to the technical field of artificial intelligence, and provides a model parameter configuration method, a device, electronic equipment and a readable storage medium, which are applied to the electronic equipment, wherein the method comprises the following steps: acquiring a model to be configured, wherein the model to be configured comprises logic modules which are divided in advance; determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics; and configuring the target preset characteristics to realize the parameter configuration of the model to be configured. The method and the device uniformly manage a plurality of preset characteristics, abstract the model to be configured in advance, divide the model into logic modules, determine the target preset characteristics for the logic models, and configure the target preset characteristics so as to realize the efficient configuration of the model to be configured.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a model parameter configuration method and device, electronic equipment and a readable storage medium.
Background
In the field of artificial intelligence technology, when a model is trained, setting of a large number of parameters is usually involved, for example, the number of parameters is large, the length of the characteristics is large, on one hand, the number of the parameters is large, on the other hand, the setting mode of each parameter is also large and complex, and when the model structure is also complex and needs to be correspondingly adjusted, a plurality of change factors are interactively accumulated, so that the workload of configuring the model in the model training process is large, and the efficiency is low.
Disclosure of Invention
The invention aims to provide a model parameter configuration method, a model parameter configuration device, an electronic device and a readable storage medium, which can simplify the configuration of a model and improve the configuration efficiency.
In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:
in a first aspect, an embodiment of the present invention provides a model parameter configuration method, which is applied to an electronic device, and the method includes: obtaining a model to be configured, wherein the model to be configured comprises logic modules which are divided in advance; determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics; and configuring the target preset characteristics to realize the parameter configuration of the model to be configured.
Optionally, a plurality of preset features are displayed on a display interface of the electronic device, and the step of determining a target preset feature from the plurality of preset features as the input of the logic module includes:
responding to a feature selection operation of a user on the display interface, wherein the feature selection operation comprises a target preset feature selected from the preset features by the user on the display interface;
and taking the target preset characteristics as input parameters of the logic module.
Optionally, the target preset feature is one, a basic configuration item of the target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset feature to implement parameter configuration of the model to be configured includes:
responding to configuration selection operation performed by a user on the display interface, wherein the configuration selection operation comprises a target configuration item selected from the basic configuration items on the display interface by the user and a configuration value of the target configuration item;
and configuring the target configuration item according to the configuration value so as to realize the parameter configuration of the model to be configured.
Optionally, the target preset features are multiple, a shared configuration item of each target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset features to implement parameter configuration of the model to be configured includes:
responding to a shared identification value of a shared configuration item of each target preset feature input on the display interface by a user, wherein the shared identification values of at least two target preset features are the same;
and establishing a sharing relation for the target preset features with the same sharing identification value, wherein the feature values of the target preset features with the sharing relation belong to the same set.
Optionally, the target preset feature includes a plurality of sub-features, a pooling configuration item of the target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset feature to implement parameter configuration of the model to be configured includes:
responding to a pooling mode of the target preset characteristics input on the display interface by a user;
and configuring the pooling configuration items according to the pooling mode to realize parameter configuration of the model to be configured.
Optionally, the method further comprises:
acquiring a characteristic value and a configuration item of a target preset characteristic of the logic module, wherein the configuration item comprises a basic configuration item and at least one of a sharing configuration item and a pooling configuration item;
and carrying out vector processing on the characteristic value according to the configuration item to obtain an output characteristic vector of the logic module.
Optionally, the target preset features are multiple, the basic configuration item of each target preset feature includes a sequential configuration item, and the step of performing vector processing on the feature values according to the configuration items to obtain the output feature vector of the logic module includes:
vectorizing the characteristic value of each target preset characteristic to obtain a characteristic vector of each target preset characteristic;
and configuring the value of the item according to the sequence of each target preset feature, and splicing the feature vectors of all the target preset features to obtain the output feature vector of the logic module.
In a second aspect, an embodiment of the present invention provides a model parameter configuration apparatus, which is applied to an electronic device, and the apparatus includes: the device comprises an acquisition module, a configuration module and a control module, wherein the acquisition module is used for acquiring a model to be configured, and the model to be configured comprises a logic module which is divided in advance; the determining module is used for determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics; and the configuration module is used for configuring the target preset characteristics so as to realize parameter configuration of the model to be configured.
In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory; the memory is used for storing programs; the processor is configured to implement the model parameter configuration method according to the first aspect when executing the program.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the model parameter configuration method as described in the first aspect.
Compared with the prior art, the model parameter configuration method, the device, the electronic equipment and the readable storage medium provided by the embodiment of the invention uniformly manage a plurality of preset features, abstract the model to be configured in advance, divide the model into the logic modules, determine the target preset features for the logic models, and configure the target preset features, so that the high-efficiency configuration of the model to be configured is realized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a diagram illustrating a logic module division of a WideDeep model according to an embodiment of the present invention.
Fig. 2 is a diagram illustrating a logic module division of the deep fm model according to an embodiment of the present invention.
Fig. 3 is a first flowchart illustrating a method for configuring model parameters according to an embodiment of the present invention.
Fig. 4 is a flowchart illustrating a method for configuring model parameters according to an embodiment of the present invention, which is shown in fig. two.
Fig. 5 is a screenshot illustrating an example of modifying operation provided in an embodiment of the present invention.
Fig. 6 is a screenshot of an exemplary operation of performing sharing configuration according to an embodiment of the present invention.
Fig. 7 is a third flowchart illustrating a method for configuring model parameters according to an embodiment of the present invention.
Fig. 8 is a block diagram schematically illustrating a model parameter configuration apparatus according to an embodiment of the present invention.
Fig. 9 is a block diagram of an electronic device according to an embodiment of the present invention.
Icon: 10-an electronic device; 11-a processor; 12-a memory; 13-a bus; 100-model parameter configuration means; 110-an obtaining module; 120-a determination module; 130-configuration module; 140-generation module.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present invention, it should be noted that if the terms "upper", "lower", "inside", "outside", etc. indicate an orientation or a positional relationship based on that shown in the drawings or that the product of the present invention is used as it is, this is only for convenience of description and simplification of the description, and it does not indicate or imply that the device or the element referred to must have a specific orientation, be constructed in a specific orientation, and be operated, and thus should not be construed as limiting the present invention.
Furthermore, the appearances of the terms "first," "second," and the like, if any, are only used to distinguish one description from another and are not to be construed as indicating or implying relative importance.
It should be noted that the features of the embodiments of the present invention may be combined with each other without conflict.
Since many parameters need to be set by an algorithm engineer during training a model, and setting items of the parameters are many and complicated, a common solution is to set the parameters in a specific format, such as json format or yaml format, and to parse the format by writing a code of a response in a model training script, which may cause the following problems:
1. whether in json format or yaml format, the formats sampled by different algorithm engineers may be different, and the non-uniformity of the formats results in the inability to multiplex the parsed code for one format with the parsed code for another format.
2. The system needs to be manually written and maintained by developers, is high in cost and is easy to make mistakes.
3. The training frames cannot be uniformly compatible, codes need to be rewritten when the model changes, and corresponding parameter processing logics need to be rewritten.
Embodiments of the present invention provide a model parameter configuration method, a model parameter configuration apparatus, an electronic device, and a readable storage medium, which are used to solve the above problems and are described in detail below.
In order to simplify the configuration of the model and improve the reusability of the features, the embodiment of the invention is improved from the following two aspects: (1) the method comprises the steps of dividing a model to be configured into one or more logic modules according to processing logic of a model structure of the model to be configured, determining input parameters of each logic module when the model to be configured is configured, configuring the input parameters, and finally realizing parameter configuration of the model to be configured. Of course, according to the training requirement, the reconfiguration of the parameters of the model to be configured can be realized by modifying the configuration of the input parameters of the logic module after the input parameters are determined. (2) The inventor analyzes the reason for frequent change of the code of model training, and finds that the features are required to be increased or decreased and the sequence of a plurality of features in a feature vector is required to be updated in the model training process, the features are required to be subjected to some conventional configurations, the same feature is repeatedly used in different logic processing stages in the same model to be configured, and the like, and the inventor abstracts the features of the model to be configured according to the analysis result so as to maximally realize the multiplexing and simplified configuration of the features.
The embodiment of the present invention first describes a partitioning of a to-be-configured model, taking a common WideDeep model as an example, please refer to fig. 1, where fig. 1 is an example diagram of logic module partitioning of the WideDeep model provided by the embodiment of the present invention, in fig. 1, a wide portion, a deep portion, and an attention portion of the WideDeep model are respectively used as a logic module, which are respectively a wide module, a deep module, and an attention module, each logic module corresponds to a respective input parameter, features ft 1-ft 4 are input parameters of the wide module, features ft 5-ft 8 are input parameters of the deep module, and features condition and ft8 are inputs of the attention module. And setting input parameters of each module to further complete parameter configuration of the whole WideDeep model. Taking another common Deep FM model as an example, fig. 2 is an illustration of logic module division of the Deep FM model provided by the embodiment of the present invention, in fig. 2, an FM1 portion, an FM2 portion, a Deep portion, and an attention portion of the Deep FM model are respectively used as a logic module, which are respectively an FM1 module, an FM2 module, a Deep module, and an attention module, features ft1, ft4, and ft7 are input parameters of the FM1 module, features ft1, ft2, ft3, ft4, and ft6 are input parameters of the FM2 module, features 1 to ft8 are input parameters of the Deep module, and features condition-fts are input parameters of the attention module.
It should be noted that the above is only an example of module division, and in fact, a user may divide the model structure according to the needs of an actual scene and the needs of logic processing in the actual scene, and finally combine the processing structures of the modules to complete the whole model structure building, thereby simplifying the building process of the model structure and reducing the development cost of the model structure building.
It should be further noted that, the user may also add a module to the model or delete an existing module through the interface, or modify the attributes of the existing module, for example, modify the name of the module.
After the model to be configured is divided, a processing procedure for configuring the logic module based on the divided model to be configured is described in detail below, please refer to fig. 3, where fig. 3 is a first flowchart of a model parameter configuration method according to an embodiment of the present invention, the method includes the following steps:
step S100, obtaining a model to be configured, wherein the model to be configured comprises logic modules which are divided in advance.
In this embodiment, the model to be configured may be a multimedia resource recommendation model, which gives a recommendation of a multimedia resource meeting a preference of a user according to a listening or watching condition of the user on the multimedia resource, where the multimedia resource may be a resource such as an audio resource, a video resource, and the like, for example, a song album, an online course, and the like.
Step S101, determining a target preset characteristic as an input parameter of the logic module from a plurality of preset characteristics.
In this embodiment, the preset features are all features related to the model to be configured, and the embodiment manages the preset features in a unified manner, extracts a plurality of parameters for the preset features, for example, each preset feature supports configuration items such as a vector size, a feature name, and a vector name, and configures the model to be configured by configuring the configuration items.
In this embodiment, the user may designate one or more preset features as the logic module input parameters, and the target preset feature is a preset feature of the plurality of preset features as the logic module input parameters. The user can conveniently appoint the incidence relation between the target preset characteristics and the logic module through the interface, development of additional codes is not needed, the user is prevented from changing codes or manually editing configuration files, the user is enabled to be concentrated on building the upper layer structure of the model, and development cost is reduced to the maximum extent.
In this embodiment, taking the multimedia resource recommendation model as an example, the preset features may be an album id listened to by the user according to a history listened to by the user, an id of a main broadcast of an album listened to by the user in a latest preset history period, a list of specific time periods listened to by the user in the latest preset history period, and the like.
And S102, configuring the target preset characteristics to realize parameter configuration of the model to be configured.
In this embodiment, as an implementation manner, a user may first specify a target preset feature for a model to be configured, and then configure the target preset feature, or may first configure the preset feature, and then specify the target preset feature for the model to be configured from the configured preset features.
In this embodiment, configuring the target preset feature may be to modify a value of a configuration item of the target preset feature, or to release a relationship between the target preset feature and the model to be configured, that is, to no longer use the target preset feature as an input parameter of the model to be configured.
According to the method provided by the embodiment, the model to be configured is abstracted in advance, the model is divided into the logic modules, the target preset characteristics are determined for the logic modules, and the target preset characteristics are configured, so that the efficient configuration of the model to be configured is realized.
Referring to fig. 4, fig. 4 is a second flowchart of a method for configuring model parameters according to an embodiment of the present invention, where step S101 includes the following sub-steps:
and a substep S1010, responding to a feature selection operation on the display interface by the user, wherein the feature selection operation comprises a target preset feature selected from a plurality of preset features on the display interface by the user.
In this implementation, in order to simplify configuration, a user may determine a target preset feature through an interface operation, an execution main body of the model parameter configuration method in this embodiment may be an electronic device, a display interface of the electronic device displays a plurality of preset features, and the user may select the target preset feature that is required to be an input parameter of a model to be configured on the display interface.
In the substep S1011, the target preset feature is used as an input parameter of the logic module.
In this embodiment, as a specific implementation manner, an association relationship between the target preset feature and the logic module may be established, and the association manner may be marked as an input parameter.
In this embodiment, the target preset feature may include different expressions, may be a single preset feature, may include a plurality of preset features, may be composed of a plurality of sub-features, and the target preset feature may be at least one of the expressions, and specific configurations may also be different in each expression.
In this embodiment, if there is one target preset feature, the manner of configuring the target preset feature may be:
firstly, responding to configuration selection operation performed by a user on a display interface, wherein the configuration selection operation comprises a target configuration item selected by the user from basic configuration items on the display interface and a configuration value of the target configuration item.
In this embodiment, the basic configuration item is an independent configuration item having a preset feature, for example, a feature length, a length of a feature vector, a feature entry and exit parameter, and the basic configuration item of the target preset feature is displayed on the display interface, and for the basic configuration item, a user may edit the basic configuration item by selecting the basic configuration item to input a configuration value of the basic configuration item, for example, the basic configuration item is: the feature name, the configuration value of the feature name is input as follows: user-id. Referring to fig. 5, fig. 5 is a screenshot illustrating an operation performed to modify according to an embodiment of the present invention. In fig. 5, the value of emb _ size, this configuration item, is modified to 10.
And secondly, configuring the target configuration item according to the configuration value so as to realize the parameter configuration of the model to be configured.
It should be noted that, when there are a plurality of target preset features and a basic configuration item of any one of the plurality of target preset features needs to be configured, the configuration may still be performed in the above configuration manner.
In this embodiment, if there are a plurality of target preset features, in order to implement sharing of the plurality of target preset features, the target preset features are configured in a shared manner, and a manner of implementing parameter configuration of the model to be configured may be:
firstly, responding to a shared identification value of a shared configuration item of each target preset feature input on a display interface by a user, wherein the shared identification values of at least two target preset features are the same.
In this embodiment, the shared configuration items of each target preset feature are displayed on the display interface, and the sharing relationship may be established by setting the same shared identification value for two or more target preset features, for example, the target preset features include features a, b, c, and d, and setting the shared identification values of the shared configuration items of b, c, and d as b-c-d means that b, c, and d have a sharing relationship.
In this embodiment, taking the multimedia resource recommendation model as an example, there may be a sharing relationship between the album id feature that the user has listened to most recently and the candidate album id feature, there may be a sharing relationship between the category feature of the album that the user has listened to most recently and the category feature of the candidate album, there may be a sharing relationship between the id feature of the anchor of the album that the user has listened to most recently and the id feature of the anchor of the candidate album, and the like.
And secondly, establishing a sharing relation for the target preset features with the same sharing identification value, wherein the feature values of the target preset features with the sharing relation belong to the same set.
In this embodiment, the target preset features having feature values belonging to the same set may be shared, the target preset features having a sharing relationship exist, if the feature values of the target preset features are the same, the finally generated feature vectors are also the same, and if the feature value of one of the target preset features changes, the other feature value having a sharing relationship also changes. Also avoiding referring to fig. 6, fig. 6 is an exemplary screenshot of an operation of performing sharing configuration according to an embodiment of the present invention, and in fig. 6, values of the feature sharing identifiers emb _ name in the 2 nd and 3 rd rows are both set to "test 1".
The feature sharing configuration provided by the embodiment can greatly simplify the maintenance workload of the shared feature, and also avoid the problems of missed modification and wrong modification which are easy to occur when the shared feature is modified.
In this embodiment, if a plurality of sub-features of the target preset feature are present, the plurality of sub-features may be merged or stacked in different pooling manners, and the target preset feature is configured in a pooling manner, so as to implement parameter configuration of the model to be configured, which may be:
firstly, responding to the pooling mode of the target preset characteristics input on the display interface by the user.
In this embodiment, the display interface displays the pooling configuration items with the target preset features, and the pooling configuration items are edited to input the pooling mode, which includes, but is not limited to, average pooling, maximum pooling, and the like.
And secondly, configuring the pooling configuration items according to a pooling mode to realize parameter configuration of the model to be configured.
It should be noted that, for a target preset feature including a plurality of sub-features, the pooling configuration items may be set or not set as required.
It should be further noted that the target preset features may include two or more of the above expression manners, and at this time, the target preset features for each manner may be configured in a corresponding manner, for example, the target preset features include features ft1 to ft5, ft1 to ft4 are independent single features, ft5 is composed of a plurality of sub-features, ft2 to ft4 are shared features, for ft1 to ft4, the basic configuration items may be configured, for ft2 to ft4, the same shared identification value may be configured for the ft2 to ft4, so as to establish a sharing relationship between the three, and for ft5, the pooling configuration items may be configured for the ft 5.
In this embodiment, the target preset feature including a plurality of sub-features is generally used to represent a behavior record of the user, taking the multimedia resource recommendation model as an example, the target preset feature may be a time period that the user listens in a preset historical period, taking a day as an example, if the user listens frequently in the evening, the time period for listening in the day may be: {17, 19, 21}, which respectively represent listening times of 17, 19, and 21 points, in which case the target preset feature includes 3 sub-features: 17. 19 and 21. Of course, the target preset feature may include a listening period of a plurality of days, and in this case, the target preset feature may include a feature of the listening period of each of the plurality of days. The target preset feature may be obtained by arranging a plurality of sub-features according to a preset sequence, where the preset sequence may be a time sequence or a position sequence.
In this embodiment, after the model to be configured is set, in order to automatically generate a feature vector according to the configuration item of the target preset feature, this embodiment further provides a specific implementation manner of generating the feature vector, please refer to fig. 7, fig. 7 is a third flowchart of the model parameter configuration method provided in this embodiment of the present invention, and the method further includes the following steps:
step S110, obtaining a characteristic value and a configuration item of a target preset characteristic of the logic module, wherein the configuration item comprises a basic configuration item and at least one of a sharing configuration item and a pooling configuration item.
In this embodiment, the basic configuration item, the sharing configuration item, and the pooling configuration item have been described in the foregoing embodiments, and are not described herein again.
And step S111, carrying out vector processing on the characteristic values according to the configuration items to obtain output characteristic vectors of the logic module.
In this embodiment, for example, when the target preset feature is one, the basic configuration item is the length of the feature vector, and the length of the feature vector is set to 10, then the length of the output feature vector after the vector processing is 10.
When the target preset features are multiple, the basic configuration items of each target preset feature further include a sequence configuration item, where the sequence configuration item is used to represent the position of a vector corresponding to the target preset feature in the output feature vector, and when the target preset features are multiple, the sequence configuration item may be configured by adjusting the arrangement sequence of the target preset features on the display interface, for example, the target preset features arranged sequentially from top to bottom on the display interface are a, b, c, and d, and the sequence may be adjusted by dragging up or down, for example, from top to bottom: a. d, b and c.
In this case, the specific implementation manner of obtaining the output feature vector of the logic module is as follows:
firstly, vectorization processing is carried out on the characteristic value of each target preset characteristic to obtain a characteristic vector of each target preset characteristic.
And secondly, configuring the values of the items according to the sequence of each target preset feature, and splicing the feature vectors of all the target preset features to obtain the output feature vector of the logic module.
For example, the target preset features arranged in sequence from top to bottom on the display interface are a, b, c and d, the corresponding vectors are emb1, emb2, emb3 and emb4, and the spliced output feature vector is emb1emb2emb3 emb 4.
When the target preset feature is composed of a plurality of sub-features and the target preset feature is configured with a pooling configuration item, a specific implementation manner of obtaining the output feature vector of the logic module is as follows:
firstly, vectorization processing is carried out on the characteristic value of each sub-characteristic to obtain a characteristic vector of each sub-characteristic.
And secondly, performing pooling treatment on the feature vectors of all the sub-features according to the pooling mode of the pooling configuration items to obtain the output feature vector of the logic module.
For example, the target preset feature ft comprises 3 sub-features ft-1, ft-2 and ft-3, the pooling mode of the pooling configuration item configuration of ft is average pooling, and the vector length is 5, then the feature vectors of ft-1, ft-2 and ft-3 are respectively expressed as: { a1a2a3a4a5}, { b1b2b3b4b5}, and { c1c2c3c4c5}, where x1 ═ avg (a1, b1, c1), x2 ═ avg (a2, b2, c2), x3 ═ avg (a3, b3, c3), x4 ═ avg (a4, b4, c4), and x5 ═ avg (a5, b5, c5), the output eigenvector is represented as { x1x2x3x4x5 }.
In addition, the embodiment may also export all configuration items into a configuration file, and preview the configuration file.
In order to execute the corresponding steps in the above embodiments and various possible implementations, an implementation of the model parameter configuration apparatus 100 is given below, which is applied to an electronic device. Referring to fig. 8, fig. 8 is a block diagram illustrating a model parameter configuration apparatus 100 according to an embodiment of the present invention. It should be noted that the basic principle and the generated technical effect of the model parameter configuration apparatus 100 provided in the present embodiment are the same as those of the above embodiments, and for the sake of brief description, no reference is made to this embodiment.
The model parameter configuration apparatus 100 includes an obtaining module 110, a determining module 120, a configuring module 130, and a generating module 140.
The obtaining module 110 is configured to obtain a model to be configured, where the model to be configured includes logic modules divided in advance.
A determining module 120, configured to determine a target preset feature from the plurality of preset features as an input parameter of the logic module.
Optionally, a plurality of preset features are displayed on a display interface of the electronic device, and the determining module 120 is specifically configured to: responding to a feature selection operation of a user on a display interface, wherein the feature selection operation comprises a target preset feature selected from a plurality of preset features by the user on the display interface; and taking the target preset characteristics as input parameters of the logic module.
The configuration module 130 is configured to configure the target preset feature to implement parameter configuration of the model to be configured.
Optionally, the target preset feature is one, a basic configuration item of the target preset feature is displayed on a display interface of the electronic device, and the configuration module 130 is specifically configured to: responding to configuration selection operation performed by a user on the display interface, wherein the configuration selection operation comprises a target configuration item selected from basic configuration items by the user on the display interface and a configuration value of the target configuration item; and configuring the target configuration item according to the configuration value so as to realize the parameter configuration of the model to be configured.
Optionally, the target preset features are multiple, a shared configuration item of each target preset feature is displayed on a display interface of the electronic device, and the configuration module 130 is further specifically configured to: responding to a shared identification value of a shared configuration item of each target preset feature input on a display interface by a user, wherein the shared identification values of at least two target preset features are the same; and establishing a sharing relation for the target preset features with the same sharing identification value, wherein the feature values of the target preset features with the sharing relation belong to the same set.
Optionally, the target preset feature includes a plurality of sub-features, a pooling configuration item of the target preset feature is displayed on a display interface of the electronic device, and the configuration module 130 is further configured to: responding to a pooling mode of target preset characteristics input on a display interface by a user; and configuring the pooling configuration items according to a pooling mode to realize parameter configuration of the model to be configured.
A generating module 140 for: acquiring a characteristic value and a configuration item of a target preset characteristic of a logic module, wherein the configuration item comprises a basic configuration item and at least one of a shared configuration item and a pooling configuration item; and carrying out vector processing on the characteristic values according to the configuration items to obtain an output characteristic vector of the logic module.
Optionally, the target preset features are multiple, the basic configuration item of each target preset feature includes a sequential configuration item, and the generating module 140 is specifically configured to: vectorizing the characteristic value of each target preset characteristic to obtain a characteristic vector of each target preset characteristic; and according to the value of the sequence configuration item of each target preset feature, splicing the feature vectors of all the target preset features to obtain the output feature vector of the logic module.
Referring to fig. 9, fig. 9 shows a block schematic diagram of the electronic device 10 according to the embodiment of the present invention, where the electronic device 10 includes a processor 11, a memory 12, and a bus 13, and the processor 11, the memory 12, and the bus 13 are connected together.
The processor 11 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the model parameter configuration method may be implemented by integrated logic circuits of hardware or instructions in the form of software in the processor 11. The Processor 11 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components.
The memory 12 is used to store a program, such as the model parameter configuration apparatus 100 in fig. 8. The model parameter configuration apparatus 100 includes at least one software functional module which may be stored in the memory 12 in the form of software or firmware (firmware) or solidified in an Operating System (OS) of the electronic device 10. After receiving the execution instruction, the processor 11 executes the program to implement the model parameter configuration method disclosed in the above embodiment.
Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the computer program implements the model parameter configuration method as described above.
In summary, embodiments of the present invention provide a model parameter configuration method, an apparatus, an electronic device, and a readable storage medium, which are applied to an electronic device, and the method includes: obtaining a model to be configured, wherein the model to be configured comprises logic modules which are divided in advance; determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics; and configuring the target preset characteristics to realize the parameter configuration of the model to be configured.
The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A model parameter configuration method is applied to electronic equipment, and the method comprises the following steps:
obtaining a model to be configured, wherein the model to be configured comprises logic modules which are divided in advance;
determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics;
and configuring the target preset characteristics to realize parameter configuration of the model to be configured.
2. The method for configuring model parameters according to claim 1, wherein a plurality of preset features are displayed on a display interface of the electronic device, and the step of determining the target preset feature as the input of the logic module from the plurality of preset features comprises:
responding to a feature selection operation of a user on the display interface, wherein the feature selection operation comprises a target preset feature selected from the plurality of preset features by the user on the display interface;
and taking the target preset characteristics as input parameters of the logic module.
3. The method for configuring model parameters according to claim 1, wherein the target preset feature is one, a basic configuration item of the target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset feature to implement parameter configuration of the model to be configured includes:
responding to configuration selection operation performed by a user on the display interface, wherein the configuration selection operation comprises a target configuration item selected from the basic configuration items on the display interface by the user and a configuration value of the target configuration item;
and configuring the target configuration item according to the configuration value so as to realize parameter configuration of the model to be configured.
4. The method for configuring model parameters according to claim 1, wherein the number of the target preset features is multiple, a shared configuration item of each target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset features to configure the parameters of the model to be configured includes:
responding to a shared identification value of a shared configuration item of each target preset feature input on the display interface by a user, wherein the shared identification values of at least two target preset features are the same;
and establishing a sharing relation for the target preset features with the same sharing identification value, wherein the feature values of the target preset features with the sharing relation belong to the same set.
5. The model parameter configuration method according to claim 1, wherein the target preset feature includes a plurality of sub-features, a pooling configuration item of the target preset feature is displayed on a display interface of the electronic device, and the step of configuring the target preset feature to configure the parameter of the model to be configured includes:
responding to a pooling mode of the target preset characteristics input on the display interface by a user;
and configuring the pooling configuration items according to the pooling mode to realize parameter configuration of the model to be configured.
6. The model parameter configuration method of claim 1, wherein the method further comprises:
acquiring a characteristic value and a configuration item of a target preset characteristic of the logic module, wherein the configuration item comprises a basic configuration item and at least one of a sharing configuration item and a pooling configuration item;
and carrying out vector processing on the characteristic value according to the configuration item to obtain an output characteristic vector of the logic module.
7. The method of claim 6, wherein the target preset features are plural, the basic configuration item of each target preset feature includes a sequential configuration item, and the step of performing vector processing on the feature values according to the configuration items to obtain the output feature vector of the logic module includes:
vectorizing the characteristic value of each target preset characteristic to obtain a characteristic vector of each target preset characteristic;
and configuring the value of the item according to the sequence of each target preset feature, and splicing the feature vectors of all the target preset features to obtain the output feature vector of the logic module.
8. A model parameter configuration device applied to electronic equipment is characterized by comprising:
the device comprises an acquisition module, a configuration module and a control module, wherein the acquisition module is used for acquiring a model to be configured, and the model to be configured comprises a logic module which is divided in advance;
the determining module is used for determining a target preset characteristic serving as an input parameter of the logic module from a plurality of preset characteristics;
and the configuration module is used for configuring the target preset characteristics so as to realize parameter configuration of the model to be configured.
9. An electronic device comprising a processor and a memory; the memory is used for storing programs; the processor is configured to implement the model parameter configuration method of any one of claims 1-7 when executing the program.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out a method for model parameter configuration according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210759448.6A CN114938332A (en) | 2022-06-29 | 2022-06-29 | Model parameter configuration method and device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210759448.6A CN114938332A (en) | 2022-06-29 | 2022-06-29 | Model parameter configuration method and device, electronic equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114938332A true CN114938332A (en) | 2022-08-23 |
Family
ID=82868866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210759448.6A Pending CN114938332A (en) | 2022-06-29 | 2022-06-29 | Model parameter configuration method and device, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114938332A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110489093A (en) * | 2019-10-16 | 2019-11-22 | 鹏城实验室 | A kind of lasting building method, device, terminal and storage medium towards AI application |
| CN111488211A (en) * | 2020-04-09 | 2020-08-04 | 北京嘀嘀无限科技发展有限公司 | Task processing method, device, equipment and medium based on deep learning framework |
| CN114444650A (en) * | 2020-11-06 | 2022-05-06 | 安霸国际有限合伙企业 | Method for improving accuracy of quantized multi-level object detection network |
| WO2022095412A1 (en) * | 2020-11-03 | 2022-05-12 | 前海飞算云智软件科技(深圳)有限公司 | Microservice development method and apparatus, storage medium, and electronic device |
-
2022
- 2022-06-29 CN CN202210759448.6A patent/CN114938332A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110489093A (en) * | 2019-10-16 | 2019-11-22 | 鹏城实验室 | A kind of lasting building method, device, terminal and storage medium towards AI application |
| CN111488211A (en) * | 2020-04-09 | 2020-08-04 | 北京嘀嘀无限科技发展有限公司 | Task processing method, device, equipment and medium based on deep learning framework |
| WO2022095412A1 (en) * | 2020-11-03 | 2022-05-12 | 前海飞算云智软件科技(深圳)有限公司 | Microservice development method and apparatus, storage medium, and electronic device |
| CN114444650A (en) * | 2020-11-06 | 2022-05-06 | 安霸国际有限合伙企业 | Method for improving accuracy of quantized multi-level object detection network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109117138B (en) | Configuration method and device and computer readable storage medium | |
| CN109754072B (en) | Processing method of network offline model, artificial intelligence processing device and related products | |
| CN109636301A (en) | Workflow configuration method, system, terminal and the readable storage medium storing program for executing of financial business | |
| CN111240774B (en) | Dynamic wallpaper generation method and device, storage medium and electronic equipment | |
| CN112733511A (en) | Form processing method, system, electronic equipment and computer readable storage medium | |
| CN112149838A (en) | Method, device, electronic equipment and storage medium for realizing automatic model building | |
| CN110633959A (en) | Method, device, equipment and medium for creating approval task based on graph structure | |
| CN111413876A (en) | Method for configuring control page of APP, cloud platform and terminal equipment | |
| CN110781180A (en) | Data screening method and data screening device | |
| CN109683948A (en) | System data introduction method, device, electronic product and storage medium | |
| CN115858078A (en) | Special effect rendering method and device, material manufacturing system, equipment and storage medium | |
| CN108804100A (en) | Create method, apparatus, storage medium and the mobile terminal of interface element | |
| CN109101311B (en) | Display view compatibility method and electronic equipment | |
| CN112044061B (en) | Game picture processing method and device, electronic equipment and storage medium | |
| CN112558968A (en) | Resource tree view generation method, device, equipment and storage medium | |
| CN114938332A (en) | Model parameter configuration method and device, electronic equipment and readable storage medium | |
| KR102593844B1 (en) | Method for constructing deep learning network, system for auto deep learning platform service and computer program for the same | |
| CN115098254A (en) | Method and system for triggering execution of subtask sequence and electronic equipment | |
| CN114281310A (en) | Page frame setting method, device, equipment, storage medium and program product | |
| CN115114554A (en) | Method and device for dragging component, terminal equipment and storage medium | |
| CN115481343A (en) | Target component generation method and device, storage medium and electronic device | |
| CN113760403A (en) | State machine linkage method and device | |
| CN108804088A (en) | protocol processing method and device | |
| CN114675819A (en) | RPA component recommendation method, device, equipment and readable storage medium | |
| CN111124386B (en) | Animation event processing method, device, equipment and storage medium based on Unity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |