Dynamic parameter adjusting method and device, electronic equipment and storage medium
By acquiring and adjusting the mapping relationship between the requested data and dynamic parameters during inference model inference, the problem of pausing inference reloading models in the prior art is solved, and the efficiency of dynamic parameter adjustment is improved.
Patent Information
- Application Number
- CN202311558092.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-21
- Publication Date
- 2025-05-23
AI Technical Summary
The existing technology cannot dynamically adjust parameters when inference model reasoning, and requires pausing the reasoning and reloading the model, resulting in inefficient adjustment.
By obtaining the request identification information and the first dynamic parameter in the parameter adjustment instruction, the second dynamic parameter is found from the mapping relationship between the request data and the dynamic parameter, and adjusting it to the first dynamic parameter to realize online reasoning.
It realizes dynamic adjustment of parameters during inference model inference, improving the efficiency of dynamic parameter adjustment.
Smart Images
Figure CN120031122A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of data processing technology, and in particular to a method and device for adjusting dynamic parameters, an electronic device, and a storage medium. Background Art
[0002] Dynamic parameters are adjustable parameters in the inference model. By adjusting the dynamic parameters, the inference data of the inference model can be optimized.
[0003] In the related technologies for adjusting dynamic parameters, the method for adjusting dynamic parameters is to adjust the dynamic parameters by reloading the configuration file and the inference model. Since the adjustment of dynamic parameters requires reloading the configuration file and the inference model, the dynamic parameters cannot be adjusted when the inference model is inferring. The inference needs to be paused and the inference model needs to be reloaded, resulting in low efficiency in adjusting the dynamic parameters. Summary of the invention
[0004] The present disclosure provides a method and device for adjusting dynamic parameters, an electronic device and a storage medium. The main purpose is to solve the problem that the adjustment efficiency of dynamic parameters is low because the dynamic parameters cannot be adjusted during inference model inference, and the inference needs to be paused and the inference model needs to be reloaded.
[0005] According to a first aspect of the present disclosure, a method for adjusting a dynamic parameter is provided, comprising:
[0006] In response to the parameter adjustment instruction, obtaining request identification information and a first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, where the first dynamic parameter is used to replace the initial parameter corresponding to the target request data;
[0007] searching, from a mapping relationship between request data and dynamic parameters, a second dynamic parameter corresponding to the request identification information, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data;
[0008] The second dynamic parameter is adjusted to the first dynamic parameter so as to perform online inference on the target request data based on the first dynamic parameter.
[0009] Optionally, the method for establishing the mapping relationship between the request data and the dynamic parameters includes:
[0010] In response to the inference request information, obtaining request data and dynamic parameters in the inference request information, wherein the dynamic parameters refer to dynamic parameters used for inference carried in the request data;
[0011] Allocating request identification information to the request data, and allocating parameter identification information to the dynamic parameter;
[0012] A binding relationship between the request identification information and the parameter identification information is established to establish a mapping relationship between the request data and the dynamic parameters.
[0013] Optionally, before obtaining, in response to the parameter adjustment instruction, the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, the method further includes:
[0014] receiving the target request data and the second dynamic parameter; wherein the target request data and the second dynamic parameter are sent by the same client;
[0015] The second dynamic parameter and the request identification information corresponding to the target request data are stored in a preset adjustment container according to a mapping relationship, so as to generate a mapping relationship of the target request data;
[0016] The target request data is forward-computationally processed to obtain first inference data.
[0017] Optionally, after performing forward calculation processing on the target request data to obtain first inference data, the method further includes:
[0018] searching, from the preset adjustment container, a first dynamic parameter corresponding to the first inference data, where the first dynamic parameter is the adjusted second dynamic parameter;
[0019] The first inference data is inferred according to the first dynamic parameter to obtain second inference data.
[0020] Optionally, the preset adjustment container is a key-value pair structure, and storing the request identification information corresponding to the second dynamic parameter and the target request data in the preset adjustment container according to a mapping relationship includes:
[0021] The request identification information corresponding to the second dynamic parameter and the target request data is stored in the preset adjustment container through a key-value pair according to a mapping relationship.
[0022] Optionally, adjusting the second dynamic parameter to the first dynamic parameter includes:
[0023] transmitting the first dynamic parameter to the preset adjustment container;
[0024] searching the preset adjustment container for a target key corresponding to the second dynamic parameter according to the request identification information;
[0025] The key value corresponding to the target key is adjusted to the key value corresponding to the first dynamic parameter.
[0026] Optionally, the method for creating the preset adjustment container includes:
[0027] In response to a configuration instruction for a preset dynamic parameter adjustment rule in the configuration file, the configuration file is configured; wherein the preset dynamic parameter adjustment rule is used to instruct to store the second dynamic parameter, and to instruct to adjust the second dynamic parameter based on the parameter adjustment instruction;
[0028] After completing the configuration of the configuration file, the preset adjustment container is created based on the configuration file.
[0029] According to a second aspect of the present disclosure, there is provided a device for adjusting a dynamic parameter, comprising:
[0030] an acquiring unit, configured to acquire, in response to a parameter adjustment instruction, request identification information and a first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, wherein the first dynamic parameter is used to replace an initial parameter corresponding to the target request data;
[0031] a searching unit, configured to search for a second dynamic parameter corresponding to the request identification information from a mapping relationship between the request data and the dynamic parameters, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data;
[0032] An adjusting unit is used to adjust the second dynamic parameter to the first dynamic parameter so as to perform online inference on the target request data based on the first dynamic parameter.
[0033] Optionally, the searching unit is further used for:
[0034] In response to the inference request information, obtaining request data and dynamic parameters in the inference request information, wherein the dynamic parameters refer to dynamic parameters used for inference carried in the request data;
[0035] Allocating request identification information to the request data, and allocating parameter identification information to the dynamic parameter;
[0036] A binding relationship between the request identification information and the parameter identification information is established to establish a mapping relationship between the request data and the dynamic parameters.
[0037] Optionally, the device further comprises:
[0038] A receiving unit, configured to receive the target request data and the second dynamic parameter; wherein the target request data and the second dynamic parameter are sent by the same client;
[0039] A storage unit, configured to store the second dynamic parameter and the request identification information corresponding to the target request data into a preset adjustment container according to a mapping relationship, and generate a mapping relationship for the target request data;
[0040] The computing unit is used to perform forward computing processing on the target request data to obtain first inference data.
[0041] Optionally, the device further comprises:
[0042] The search unit is further used to search the first dynamic parameter corresponding to the first inference data from the preset adjustment container, where the first dynamic parameter is the adjusted second dynamic parameter;
[0043] An inference unit is used to perform inference processing on the first inference data according to the first dynamic parameter to obtain second inference data.
[0044] Optionally, the storage unit is further used for:
[0045] The request identification information corresponding to the second dynamic parameter and the target request data is stored in the preset adjustment container through a key-value pair according to a mapping relationship.
[0046] Optionally, the adjustment unit includes:
[0047] A transmission module, used for transmitting the first dynamic parameter to the preset adjustment container;
[0048] A search module, configured to search the preset adjustment container for a target key corresponding to the second dynamic parameter according to the request identification information;
[0049] The adjustment module is used to adjust the key value corresponding to the target key to the key value corresponding to the first dynamic parameter.
[0050] Optionally, the storage unit is further used for:
[0051] In response to a configuration instruction for a preset dynamic parameter adjustment rule in the configuration file, the configuration file is configured; wherein the preset dynamic parameter adjustment rule is used to instruct to store the second dynamic parameter, and to instruct to adjust the second dynamic parameter based on the parameter adjustment instruction;
[0052] After completing the configuration of the configuration file, the preset adjustment container is created based on the configuration file.
[0053] According to a third aspect of the present disclosure, there is provided an electronic device, including:
[0054] at least one processor; and
[0055] a memory communicatively connected to the at least one processor; wherein,
[0056] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.
[0057] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect.
[0058] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method as described in the first aspect above.
[0059] The dynamic parameter adjustment method and device, electronic device and storage medium provided by the present disclosure, in response to the parameter adjustment instruction, obtain the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, the first dynamic parameter is used to replace the initial parameter corresponding to the target request data; from the mapping relationship between the request data and the dynamic parameter, find the second dynamic parameter corresponding to the request identification information, wherein the mapping relationship is the mapping relationship between the request data and the dynamic parameter established after receiving the request data, and the second dynamic parameter is the initial parameter corresponding to the target request data; adjust the second dynamic parameter to the first dynamic parameter, so as to perform online reasoning on the target request data based on the first dynamic parameter. Compared with the related art, the embodiment of the present application obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, finds the second dynamic parameter corresponding to the identification information, and adjusts the second dynamic parameter to the first dynamic parameter, so that the dynamic parameter is adjusted during the reasoning of the reasoning model, thereby improving the adjustment efficiency of the dynamic parameter.
[0060] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.
[0062] Figure 1 A schematic diagram of a flow chart of a method for adjusting dynamic parameters provided by an embodiment of the present disclosure;
[0063] Figure 2 A schematic diagram of the architecture of an inference system provided by an embodiment of the present disclosure;
[0064] Figure 3 A flowchart of a method for establishing a mapping relationship between request data and dynamic parameters provided in an embodiment of the present disclosure;
[0065] Figure 4 A schematic diagram of the structure of a dynamic parameter adjustment device provided by an embodiment of the present disclosure;
[0066] Figure 5 A schematic diagram of the structure of another dynamic parameter adjustment device provided by an embodiment of the present disclosure;
[0067] Figure 6 A schematic block diagram of an exemplary electronic device provided for an embodiment of the present disclosure. DETAILED DESCRIPTION
[0068] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.
[0069] The following describes a method and apparatus for adjusting dynamic parameters, an electronic device, and a storage medium according to an embodiment of the present disclosure with reference to the accompanying drawings.
[0070] Figure 1 A flowchart of a method for adjusting dynamic parameters provided in an embodiment of the present disclosure is provided.
[0071] like Figure 1 As shown, the method is applied to a processor of a server, and the method comprises the following steps:
[0072] Step 101, in response to a parameter adjustment instruction, obtaining request identification information and a first dynamic parameter corresponding to target request data carried in the parameter adjustment instruction, wherein the first dynamic parameter is used to replace an initial parameter corresponding to the target request data.
[0073] In order to better understand the adjustment of dynamic parameters, such as Figure 2 As shown, Figure 2The present invention is a schematic diagram of the architecture of an inference system provided by an embodiment of the present invention; when the inference model infers the target request data, the dynamic parameters are required, wherein the inference model generates coherent and meaningful text content through given request data and dynamic parameters, and in order to make the inference model infer better, the dynamic parameters may need to be adjusted. From the machine level, when adjusting the dynamic parameters, from the human perspective, the user inputs the first dynamic parameter in the operation interface and triggers the adjustment button. From the machine level, in response to the parameter adjustment instruction sent to the inference system (the parameter adjustment instruction is generated after the user triggers the adjustment button), the inference system parses the parameter adjustment instruction and obtains the request identification information and the first dynamic parameter corresponding to the target request data in the parameter adjustment instruction, wherein the target request data is the request data corresponding to the dynamic parameter to be adjusted, the request identification information is the identification information of the target request data, the target request data can be determined from a large number of request data through the request identification information, the first dynamic parameter is used to replace the initial parameter corresponding to the target request data, and the second dynamic parameter corresponding to the target request data is adjusted to the first dynamic parameter, and the second dynamic parameter is the initial parameter corresponding to the target request data, so that the inference model can achieve better inference effect.
[0074] Step 102: Search for a second dynamic parameter corresponding to the request identification information from a mapping relationship between the request data and the dynamic parameters, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data.
[0075] Please continue reading Figure 2 When adjusting the dynamic parameters, it is necessary to find the second dynamic parameter of the target request data, the second dynamic parameter is the initial parameter corresponding to the target request data, the second dynamic parameter and the target request data are sent to the reasoning system together by the client, the reasoning system establishes a mapping relationship between the target request data and the second dynamic parameter in the preset dynamic library, and aggregates the target request data and the second dynamic parameter through the preset dynamic library, and transmits the aggregated data to the online reasoning engine for reasoning, wherein the preset dynamic library is generated by a preset backend code based on a preset reasoning framework, the target request data corresponding to the request identification information obtained in step 101 can be used to determine the target request data, and the second dynamic parameter can be queried and obtained according to the mapping relationship between the target request data and the second dynamic parameter.
[0076] Step 103: Adjust the second dynamic parameter to the first dynamic parameter, so as to perform online inference on the target request data based on the first dynamic parameter.
[0077] Please continue to refer to Figure 2 After the target request data and the second dynamic parameter are transmitted to the online inference engine, the second dynamic parameter is stored in a preset adjustment container. The target request data is calculated through a preset decoding module of the inference model to convert the target request data into data recognizable by the inference model. This process does not require the use of the second dynamic parameter. When the preset decoding module transmits the converted data recognizable by the inference model to the preset search module of the inference model for calculation, the preset search module obtains the second dynamic parameter from the preset adjustment container through the mapping relationship between the target request data and the second dynamic parameter for inference. Therefore, by adjusting the second dynamic parameter to the first dynamic parameter, the inference model can perform online inference on the target request data based on the first dynamic parameter.
[0078] It should be noted that the preset inference framework and the online inference engine can be located in the same carrier or in different carriers, which is not limited in this embodiment.
[0079] The method for adjusting the dynamic parameter provided by the present disclosure responds to a parameter adjustment instruction, obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, where the first dynamic parameter is used to replace the initial parameter corresponding to the target request data; searches for the second dynamic parameter corresponding to the request identification information from the mapping relationship between the request data and the dynamic parameter, where the mapping relationship is the mapping relationship between the request data and the dynamic parameter established after receiving the request data, and the second dynamic parameter is the initial parameter corresponding to the target request data; adjusts the second dynamic parameter to the first dynamic parameter so as to perform online inference on the target request data based on the first dynamic parameter. Compared with the related art, the embodiment of the present application obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, searches for the second dynamic parameter corresponding to the identification information, and adjusts the second dynamic parameter to the first dynamic parameter, so that the dynamic parameter is adjusted during the inference of the inference model, improving the adjustment efficiency of the dynamic parameter.
[0080] As a refinement of step 102, when implementing the method for establishing the mapping relationship between the request data and the dynamic parameter, it can be implemented by, but not limited to, the following methods, such as Figure 3 shown Figure 3 FIG. is a schematic flowchart of a method for establishing a mapping relationship between request data and dynamic parameters provided by an embodiment of the present disclosure, including:
[0081] Step 201, in response to the inference request information, obtain the request data and dynamic parameters in the inference request information, wherein the dynamic parameters refer to the dynamic parameters used for inference carried in the request data.
[0082] The client sends an inference request message to the inference system, wherein the inference request message includes request data and dynamic parameters, and the dynamic parameters refer to the dynamic parameters used for inference carried in the request data. After receiving the inference request message, the inference system parses the inference request message, obtains the request data and dynamic parameters in the inference request message, and transmits the request information and dynamic parameters to the preset dynamic library, allocates request identification information to the request data, and prepares for allocating parameter identification information to the dynamic parameters.
[0083] Step 202: assign request identification information to the request data, and assign parameter identification information to the dynamic parameter.
[0084] In the preset dynamic library, the preset dynamic library assigns request identification information to the request data, and assigns parameter identification information to the dynamic parameters, wherein the request identification information is the identification information of the request data, through which the corresponding request data can be determined, and the parameter identification information is the identification information of the dynamic parameters, through which the corresponding dynamic parameters can be determined, in preparation for establishing a mapping relationship between the request data and the dynamic parameters.
[0085] Step 203: Establish a binding relationship between the request identification information and the parameter identification information to establish a mapping relationship between the request data and the dynamic parameters.
[0086] Different clients will send request data and dynamic parameters. In order to enable the inference model to query the dynamic parameters corresponding to the request data when performing online inference on the request data, where the dynamic parameters corresponding to the request data are dynamic parameters from the same client as the request data, it is necessary to establish a mapping relationship between the request data and the dynamic parameters from the same sending source. By establishing a binding relationship between the request identification information corresponding to the request data and the parameter identification information corresponding to the dynamic parameters, and then determining the mapping relationship between the request data and the dynamic parameters, it can be ensured that the request data can query the corresponding dynamic parameters. For example, request data 1 and dynamic parameter 2 sent by client a are received, and request data 3 and dynamic parameter 4 sent by client b are received. By binding the request identification information of request data 1 with the parameter identification information of dynamic parameter 2, the request identification information of request data 3 is bound to the parameter identification information of dynamic parameter 4.
[0087] In actual applications, before responding to a parameter adjustment instruction and obtaining the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, the step of the inference model performing online inference on the target request data is divided into two steps. The first step is to decode the request data to obtain the first inference data, and the second step is to perform further online inference based on the first inference data and the second dynamic parameter to obtain the second inference data, wherein the first step can be implemented in but not limited to the following manner: receiving the target request data and the second dynamic parameter; wherein the target request data and the second dynamic parameter are sent by the same client; storing the request identification information corresponding to the second dynamic parameter and the target request data in a preset adjustment container according to a mapping relationship, generating a mapping relationship of the target request data, and preparing for the adjustment of the second dynamic parameter or the online inference of the second step; decoding the target request data through forward calculation, wherein the forward calculation is a basic calculation process in the inference model, and in the forward calculation process, the target request data is transmitted layer by layer through the preset decoding module of the inference model to finally obtain the output result; converting the target request data into data recognizable by the inference model to obtain the first inference data.
[0088] In actual applications, after the target request data is forward calculated and processed to obtain the first inference data, the first step of the online inference model of the target request data is completed. The second step of the online inference model of the target request data can be implemented in the following way but is not limited to: searching the first dynamic parameter corresponding to the first inference data from the preset adjustment container, where the first dynamic parameter is the adjusted second dynamic parameter; inferring the first inference data according to the first dynamic parameter by calling the preset search module to obtain the second inference data, wherein the first dynamic parameter is used to limit the output result of the preset search module, for example, the inference result of the preset search module for the first inference data is 50 words, and the limited range of the first dynamic parameter is 25 words. The preset search module will re-perform online inference based on the first dynamic parameter to obtain an inference result of 25 words; if the inference system receives a parameter adjustment instruction, it will adjust the first dynamic parameter.
[0089] As a refinement of the above embodiment, the preset adjustment container is a key-value pair structure. When executing the storage of the request identification information corresponding to the second dynamic parameter and the target request data in the preset adjustment container according to the mapping relationship, the second dynamic parameter and the request identification information corresponding to the target request data are stored in the preset adjustment container through a key-value pair according to the mapping relationship, wherein each key can only store a pair of request identification information corresponding to the second dynamic parameter and the target request data, for example, key A stores dynamic parameter 2 and request identification information c corresponding to request data 1, and key B stores dynamic parameter 4 and request identification information d corresponding to request data 3.
[0090] As a refinement of the above embodiment, when adjusting the second dynamic parameter to the first dynamic parameter, the dynamic parameter is adjusted through the preset adjustment container. There may be multiple dynamic parameters in the preset adjustment container. In order to ensure that the second dynamic parameter can be accurately adjusted to the first dynamic parameter, it can be implemented in but not limited to the following manner: transmitting the first dynamic parameter to the preset adjustment container; searching the target key corresponding to the second dynamic parameter in the preset adjustment container according to the request identification information; adjusting the key value corresponding to the target key to the key value corresponding to the first dynamic parameter, wherein the key value is the storage form of the dynamic parameter in the preset adjustment container.
[0091] As a refinement of the above embodiment, when executing the method for creating the preset adjustment container, the configuration file of the preset inference architecture is configured according to the configuration instructions of the preset dynamic parameter adjustment rules in the configuration file. After completing the configuration of the configuration file, the configured configuration file is compiled by an encoder based on the preset back-end code of the preset inference framework, and the configured configuration file is compiled into a preset adjustment container, wherein the preset dynamic parameter adjustment rule is used to indicate the storage of the second dynamic parameter, and to indicate the adjustment of the second dynamic parameter based on the parameter adjustment instruction. However, it should be clear that this statement is not intended to limit the method for creating the preset dynamic library.
[0092] In summary, the embodiments of the present disclosure can achieve the following effects:
[0093] 1. The embodiment of the present disclosure obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, searches for the second dynamic parameter corresponding to the identification information, adjusts the second dynamic parameter to the first dynamic parameter, and adjusts the dynamic parameter during the inference of the inference model, thereby improving the adjustment efficiency of the dynamic parameter.
[0094] Corresponding to the above-mentioned method for adjusting dynamic parameters, the present invention also provides a device for adjusting dynamic parameters. Since the device embodiment of the present invention corresponds to the above-mentioned method embodiment, details not disclosed in the device embodiment can be referred to the above-mentioned method embodiment, and will not be described in detail in the present invention.
[0095] Figure 4 A structural diagram of a dynamic parameter adjustment device provided in an embodiment of the present disclosure is shown in FIG. Figure 4 As shown, including:
[0096] An acquiring unit 31 is configured to acquire, in response to a parameter adjustment instruction, request identification information and a first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, wherein the first dynamic parameter is used to replace an initial parameter corresponding to the target request data;
[0097] A searching unit 32, configured to search for a second dynamic parameter corresponding to the request identification information from a mapping relationship between the request data and the dynamic parameters, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data;
[0098] The adjusting unit 33 is configured to adjust the second dynamic parameter to the first dynamic parameter so as to perform online inference on the target request data based on the first dynamic parameter.
[0099] The dynamic parameter adjustment device provided by the present disclosure, in response to the parameter adjustment instruction, obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, the first dynamic parameter is used to replace the initial parameter corresponding to the target request data; from the mapping relationship between the request data and the dynamic parameter, finds the second dynamic parameter corresponding to the request identification information, wherein the mapping relationship is the mapping relationship between the request data and the dynamic parameter established after receiving the request data, and the second dynamic parameter is the initial parameter corresponding to the target request data; adjusts the second dynamic parameter to the first dynamic parameter, so as to perform online reasoning on the target request data based on the first dynamic parameter. Compared with the related art, the embodiment of the present application obtains the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, finds the second dynamic parameter corresponding to the identification information, and adjusts the second dynamic parameter to the first dynamic parameter, so that the dynamic parameter is adjusted during the reasoning of the reasoning model, thereby improving the adjustment efficiency of the dynamic parameter.
[0100] Furthermore, in a possible implementation of this embodiment, Figure 5 As shown, the search unit 32 is also used for:
[0101] In response to the inference request information, obtaining request data and dynamic parameters in the inference request information, wherein the dynamic parameters refer to dynamic parameters used for inference carried in the request data;
[0102] Allocating request identification information to the request data, and allocating parameter identification information to the dynamic parameter;
[0103] A binding relationship between the request identification information and the parameter identification information is established to establish a mapping relationship between the request data and the dynamic parameters.
[0104] Furthermore, in a possible implementation of this embodiment, Figure 5 As shown, the device also includes:
[0105] The receiving unit 34 is used to receive the target request data and the second dynamic parameter; wherein the target request data and the second dynamic parameter are sent by the same client;
[0106] The storage unit 35 is used to store the second dynamic parameter and the request identification information corresponding to the target request data into a preset adjustment container according to a mapping relationship, so as to generate a mapping relationship of the target request data;
[0107] The computing unit 36 is used to perform forward computing processing on the target request data to obtain first inference data.
[0108] Furthermore, in a possible implementation of this embodiment, Figure 5 As shown, the device also includes:
[0109] The search unit 32 is further used to search the first dynamic parameter corresponding to the first inference data from the preset adjustment container, where the first dynamic parameter is the adjusted second dynamic parameter;
[0110] The inference unit 37 is used to perform inference processing on the first inference data according to the first dynamic parameter to obtain second inference data.
[0111] Furthermore, in a possible implementation of this embodiment, Figure 5 As shown, the storage unit 35 is also used for:
[0112] The request identification information corresponding to the second dynamic parameter and the target request data is stored in the preset adjustment container through a key-value pair according to a mapping relationship.
[0113] Furthermore, in a possible implementation of this embodiment, Figure 5 As shown, the adjustment unit 33 includes:
[0114] A transmission module 331, configured to transmit the first dynamic parameter to the preset adjustment container;
[0115] A search module 332, configured to search the preset adjustment container for a target key corresponding to the second dynamic parameter according to the request identification information;
[0116] The adjustment module 333 is used to adjust the key value corresponding to the target key to the key value corresponding to the first dynamic parameter.
[0117] Furthermore, in a possible implementation of this embodiment, as Figure 5 As shown, the storage unit 35 is also used for:
[0118] In response to a configuration instruction for a preset dynamic parameter adjustment rule in the configuration file, the configuration file is configured; wherein the preset dynamic parameter adjustment rule is used to instruct to store the second dynamic parameter, and to instruct to adjust the second dynamic parameter based on the parameter adjustment instruction;
[0119] After completing the configuration of the configuration file, the preset adjustment container is created based on the configuration file.
[0120] It should be noted that the above explanation of the method embodiment is also applicable to the device of this embodiment, and the principle is the same, which is not limited in this embodiment.
[0121] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.
[0122] Figure 6 A schematic block diagram of an example electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.
[0123] like Figure 6As shown, the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 402 or a computer program loaded from a storage unit 408 to a RAM (Random Access Memory) 403. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An I / O (Input / Output) interface 405 is also connected to the bus 404.
[0124] A number of components in the device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0125] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as a method for adjusting dynamic parameters. For example, in some embodiments, the method for adjusting dynamic parameters may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to execute the aforementioned dynamic parameter adjustment method in any other appropriate manner (for example, by means of firmware).
[0126] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application Specific Standard Products), SOCs (System On Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor that may be a dedicated or general-purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0127] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.
[0128] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a RAM, a ROM, an EPROM (Electrically Programmable Read-Only-Memory) or a flash memory, an optical fiber, a CD-ROM (Compact Dis sc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0129] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).
[0130] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.
[0131] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.
[0132] It should be noted that artificial intelligence is a discipline that studies how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), and includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, as well as machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.
[0133] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.
[0134] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.
Claims
1. A method for adjusting dynamic parameters, It is characterized in that include: In response to the parameter adjustment instruction, obtaining request identification information and a first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, where the first dynamic parameter is used to replace the initial parameter corresponding to the target request data; searching, from a mapping relationship between request data and dynamic parameters, a second dynamic parameter corresponding to the request identification information, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data; The second dynamic parameter is adjusted to the first dynamic parameter, so as to perform online inference on the target request data based on the first dynamic parameter.
2. The method according to claim 1, It is characterized in that The method for establishing the mapping relationship between the request data and the dynamic parameters includes: In response to the inference request information, obtaining request data and dynamic parameters in the inference request information, wherein the dynamic parameters refer to dynamic parameters used for inference carried in the request data; Allocating request identification information to the request data, and allocating parameter identification information to the dynamic parameter; A binding relationship between the request identification information and the parameter identification information is established to establish a mapping relationship between the request data and the dynamic parameters.
3. The method according to claim 1, It is characterized in that Before obtaining, in response to the parameter adjustment instruction, the request identification information and the first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, the method further includes: Receive target request data and a second dynamic parameter; wherein the target request data and the second dynamic parameter are sent by the same client; The second dynamic parameter and the request identification information corresponding to the target request data are stored in a preset adjustment container according to a mapping relationship, so as to generate a mapping relationship of the target request data; The target request data is forward-computationally processed to obtain first inference data.
4. The method according to claim 3, It is characterized in that After forward computing the target request data to obtain first inference data, the method further includes: searching, from the preset adjustment container, a first dynamic parameter corresponding to the first inference data, where the first dynamic parameter is the adjusted second dynamic parameter; The first inference data is inferred according to the first dynamic parameter to obtain second inference data.
5. The method according to claim 3, It is characterized in that The preset adjustment container is a key-value pair structure, and storing the request identification information corresponding to the second dynamic parameter and the target request data into the preset adjustment container according to a mapping relationship includes: The request identification information corresponding to the second dynamic parameter and the target request data is stored in the preset adjustment container through a key-value pair according to a mapping relationship.
6. The method according to claim 3, It is characterized in that The adjusting the second dynamic parameter to the first dynamic parameter comprises: transmitting the first dynamic parameter to the preset adjustment container; searching the preset adjustment container for a target key corresponding to the second dynamic parameter according to the request identification information; The key value corresponding to the target key is adjusted to the key value corresponding to the first dynamic parameter.
7. The method according to any one of claims 3 to 6, It is characterized in that The method for creating the preset adjustment container includes: In response to a configuration instruction for a preset dynamic parameter adjustment rule in the configuration file, the configuration file is configured; wherein the preset dynamic parameter adjustment rule is used to instruct to store the second dynamic parameter, and to instruct to adjust the second dynamic parameter based on the parameter adjustment instruction; After completing the configuration of the configuration file, the preset adjustment container is created based on the configuration file.
8. A device for adjusting dynamic parameters, It is characterized in that include: an acquiring unit, configured to acquire, in response to a parameter adjustment instruction, request identification information and a first dynamic parameter corresponding to the target request data carried in the parameter adjustment instruction, wherein the first dynamic parameter is used to replace an initial parameter corresponding to the target request data; a searching unit, configured to search for a second dynamic parameter corresponding to the request identification information from a mapping relationship between the request data and the dynamic parameters, wherein the mapping relationship is a mapping relationship between the request data and the dynamic parameters established after receiving the request data, and the second dynamic parameter is an initial parameter corresponding to the target request data; An adjusting unit is used to adjust the second dynamic parameter to the first dynamic parameter so as to perform online inference on the target request data based on the first dynamic parameter.
9. An electronic device, It is characterized in that include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium storing computer instructions, It is characterized in that The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.