Model training method, device, equipment, vehicle and computer program product
By constructing a training environment and outputting safety warnings in the vehicle's intelligent cockpit system, the problem of high computing power requirements for large model training is solved, and safe and efficient model training is achieved.
Patent Information
- Application Number
- CN202511136731.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-13
- Publication Date
- 2025-11-11
AI Technical Summary
The training process for large models requires high computing power from the platform, making them unusable directly and potentially affecting the normal operation of vehicles.
The training environment for the target model is constructed using the vehicle's intelligent cockpit system. The model is trained according to the training configuration parameters and limiting configuration parameters. When the vehicle's operating status does not meet the requirements, a safety warning is output to ensure that the training does not affect the normal operation of the vehicle.
It meets the computing power requirements for training large models while avoiding the impact of training on the normal operation of the vehicle, thus ensuring driving safety.
Smart Images
Figure CN120930706A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of model training technology, and more particularly to model training methods, apparatus, equipment, vehicles, and computer program products. Background Technology
[0002] In related technical fields, large-scale models are receiving increasing attention and are being widely used. Using large-scale models can bring many conveniences to people's lives; however, some large-scale models cannot be used directly and require training or fine-tuning, a process that places high demands on the platform's computing power. Therefore, how to meet the computing power requirements for training large-scale models has become a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0003] To address or partially address the problems existing in related technologies, this application provides a model training method, apparatus, equipment, vehicle, and computer program product that can meet the computing power requirements for training large models.
[0004] The first aspect of this application provides a method for training a model, applied to a vehicle's intelligent cockpit system, the method comprising: In response to the training instructions of the target model, construct the training environment required for the target model; In the training environment, the target model is trained according to the model training configuration parameters and the model constraint configuration parameters; the model training configuration parameters include the configuration information of the target model and the training data of the target model, and the model constraint configuration parameters include the configuration information of the resources available to the target model during the training process; During training, if the vehicle's operating status is detected to be inconsistent with the training requirements of the target model, a safety warning is output.
[0005] Furthermore, in the method described above, constructing the training environment required for the target model in response to the training instruction of the target model includes: In response to the training instructions for the target model, the training environment required for the target model is constructed using Docker or a virtual machine.
[0006] Furthermore, the method described above also includes: During the training process of the target model, the training information of the target model is displayed; the training information includes at least one of the following: training time, training progress, current computing power used, and current memory used.
[0007] Furthermore, in the method described above, the step of outputting a safety warning during the training process if the vehicle's operating state is detected to be inconsistent with the training requirements of the target model includes: If the vehicle is detected to have exited the parking position or the vehicle's battery level is detected to be less than a first preset threshold, a first safety warning is issued; if the vehicle's battery level is detected to be less than a second preset threshold, a second safety warning is issued and the training process of the target model is stopped; the first preset threshold is greater than the second preset threshold. The first security alert is output via message push, and the second security alert is output via telephone reminder.
[0008] Furthermore, the method described above also includes: In response to a usage instruction from the in-vehicle local model, the in-vehicle local model is run; the usage instruction carries target data that needs to be processed by the in-vehicle local model, and the sender of the usage instruction includes a user terminal that is communicatively connected to the intelligent cockpit system. The processing results of the target data by the in-vehicle local model are fed back to the user terminal.
[0009] Furthermore, in the method described above, the model training configuration parameters include the download path of the training data; The download path for the training data includes downloading the training data from a storage device connected to the intelligent cockpit system; the storage device establishes a connection with the intelligent cockpit system via a network hotspot, the cloud, or by accessing the web interface of the intelligent cockpit system.
[0010] A second aspect of this application provides a model training device for use in a vehicle's intelligent cockpit system, the device comprising: A construction module is used to construct the training environment required by the target model in response to the training instructions of the target model; A training module is used to train the target model in the training environment according to model training configuration parameters and model constraint configuration parameters; the model training configuration parameters include configuration information of the target model and the training data of the target model, and the model constraint configuration parameters include configuration information of the resources available to the target model during training; The early warning module is used to output a safety warning if the vehicle's operating status does not meet the training requirements of the target model during the training process.
[0011] A third aspect of this application provides an electronic device, comprising: Processor; and A memory that stores executable code, which, when executed by the processor, causes the processor to perform the method described above.
[0012] A fourth aspect of this application provides a vehicle, comprising: Intelligent cockpit system; The intelligent cockpit system is configured to implement the methods described above.
[0013] The fifth aspect of this application provides a computer program product comprising computer instructions that, when executed by a processor, implement the method described above.
[0014] The technical solution provided in this application may include the following beneficial results: The technical solution of this application is applied to an intelligent cockpit system for vehicles. The intelligent cockpit system responds to training instructions from a target model, constructs the training environment required by the target model, and trains the target model according to model training configuration parameters and model constraint configuration parameters. The model training configuration parameters include configuration information for the target model and its training data, while the model constraint configuration parameters include configuration information for available resources during training. During training, if the vehicle's operating state is detected as not meeting the training requirements of the target model, a safety warning is output. This configuration enables the intelligent cockpit system to train the target model, thus meeting the computational power requirements for model training.
[0015] Furthermore, the intelligent cockpit system trains the target model according to the model training configuration parameters and model constraint configuration parameters, and outputs a safety warning when it detects that the vehicle's operating state does not meet the training requirements of the target model. This can effectively avoid the impact of training the target model on the normal operation of the vehicle and ensure the vehicle's driving safety.
[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0017] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.
[0018] Figure 1 This is a feasible application scenario for the training method of the model shown in the embodiments of this application; Figure 2 This is a schematic flowchart illustrating a model training method according to an embodiment of this application; Figure 3 This is a flowchart illustrating another model training method according to an embodiment of this application; Figure 4 This is a signaling diagram illustrating the usage process of the vehicle-mounted local model as shown in the embodiments of this application; Figure 5This is a schematic diagram of the structure of the training device for the model shown in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of an electronic device shown in an embodiment of this application; Figure 7 This is a schematic diagram of the vehicle structure shown in the embodiments of this application. Detailed Implementation
[0019] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.
[0020] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.
[0021] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0022] In related technological fields, large models are receiving increasing attention and are being widely applied. These models, with their powerful generalization and multi-task processing capabilities, have demonstrated transformative potential in multiple fields such as natural language processing and computer vision. Using large models can bring numerous conveniences to people's lives, such as improving service efficiency through intelligent customer service, optimizing treatment processes with AI-assisted diagnosis in healthcare, and improving user experience through personalized recommendation systems.
[0023] While using large models can bring many conveniences to people's lives, some large models cannot be used directly and require training or fine-tuning, a process that places high demands on the platform's computing power. Therefore, how to meet the computing power requirements for training large models has become an urgent problem to be solved by those skilled in the art.
[0024] To address the aforementioned issues, embodiments of this application provide a model training method, apparatus, device, vehicle, and computer program product that can use an intelligent cockpit system to train a target model, thereby meeting the computing power required for model training and avoiding the impact of training the target model on the normal operation of the vehicle, thus ensuring vehicle driving safety.
[0025] The technical solutions of the embodiments of this application are described in detail below with reference to the accompanying drawings.
[0026] Figure 1 The feasible application scenarios of the model training method are shown, such as... Figure 1 The scenario shown includes a vehicle's intelligent cockpit system, a user terminal, and a cloud server.
[0027] The aforementioned vehicles refer to motor vehicles, which can be either new energy vehicles or conventional energy vehicles; this embodiment does not impose any limitations. The vehicle's intelligent cockpit system utilizes sensors, AI algorithms, and in-vehicle connectivity technologies to achieve environmental perception both inside and outside the vehicle, human-machine interaction, and customized services.
[0028] The user terminal can be an electronic device with network access capabilities. Specifically, for example, the user terminal can be a desktop computer, tablet computer, laptop computer, smartphone, digital assistant, smart wearable device, shopping guide terminal, television set, etc. Among them, smart wearable devices include, but are not limited to, smart bracelets, smartwatches, smart glasses, smart helmets, smart necklaces, etc. Alternatively, the user terminal can also be software that can run on an electronic device.
[0029] A cloud server can be an electronic device with a certain computing power. It may include a network communication module, a processor, and memory. Of course, a cloud server can also refer to software running on an electronic device. A cloud server can also be a distributed server, a system with multiple processors, memory, network communication modules, etc., working collaboratively. Alternatively, a cloud server can be a server cluster formed by several servers. Furthermore, with the development of science and technology, a cloud server can also be a new technological means capable of realizing the corresponding functions of the embodiments described in the specification. For example, it could be a new form of "server" based on quantum computing.
[0030] Both the user terminal and the intelligent cockpit system communicate with the cloud server. Specifically, the user terminal communicates with the cloud server through a first target network, and the intelligent cockpit system communicates with the cloud server through a second target network. The first and second target networks can be of any type.
[0031] For example, the first target network can be a single network, further subdivided into multiple sub-networks. Specifically, the first target network or the multiple sub-networks it comprises can be at least one of cellular mobile networks (e.g., 2G, 3G, 4G, or 5G), ZigBee, Wi-Fi, and Bluetooth, or any combination of at least one of these networks with other networks. Similarly, the second target network can be a single network, further subdivided into multiple sub-networks. Specifically, the second target network or the multiple sub-networks it comprises can be at least one of cellular mobile networks (e.g., 2G, 3G, 4G, or 5G), ZigBee, Wi-Fi, and Bluetooth, or any combination of at least one of these networks with other networks.
[0032] Based on the above feasible application scenarios, users can send training instructions for the target model through the user terminal, and the cloud server will receive the training instructions and send them to the intelligent cockpit system. Alternatively, users can directly send training instructions for the target model to the intelligent cockpit system through the human-computer interaction device on the intelligent cockpit system.
[0033] The vehicle's intelligent cockpit system responds to the training instructions of the target model and constructs the training environment required by the target model. In the training environment, the target model is trained according to the model training configuration parameters and model constraint configuration parameters. The model training configuration parameters include the configuration information of the target model and its training data, and the model constraint configuration parameters include the configuration information of the resources available to the target model during training. During training, if the vehicle's operating status is detected to be inconsistent with the training requirements of the target model, a safety warning is output.
[0034] The cloud server receives the security alert and sends it to the user's client to inform the user.
[0035] This setup allows the intelligent cockpit system to train the target model, meeting the computing power requirements for model training. Furthermore, the intelligent cockpit system trains the target model according to the model training configuration parameters and model limitation configuration parameters, and outputs a safety warning when it detects that the vehicle's operating state does not meet the training requirements of the target model. This effectively prevents the training of the target model from affecting the normal operation of the vehicle, ensuring driving safety.
[0036] Furthermore, this application proposes a model training method, which is applied to a vehicle's intelligent cockpit system. See also... Figure 2 As shown, the method includes: S101. In response to the training instructions of the target model, construct the training environment required by the target model.
[0037] The target model refers to the model that needs to be trained. In the embodiments of this application, the type of target model is not limited. The target model can be any model, such as a regular neural network model or a large language model (LLM).
[0038] The training instructions for the target model can come from the human-computer interaction device of the intelligent cockpit system. That is, the user can send the training instructions for the target model to the intelligent cockpit system based on the human-computer interaction device of the intelligent cockpit system. The human-computer interaction device can be a button, touch screen or voice interaction device of the intelligent cockpit system, etc., and this embodiment does not limit it.
[0039] The training instructions for the target model can also come from the user terminal. That is, the user can send the training instructions for the target model through the user terminal, and then the cloud server will obtain the training instructions and send them to the intelligent cockpit system.
[0040] In response to the training instructions of the target model, the vehicle's intelligent cockpit system constructs the training environment required for the target model. It should be noted that the training environment required for the target model can be a general environment for model training; that is, regardless of the type of the target model, the constructed training environment can be a general environment for model training.
[0041] Specifically, the training environment required for model training can be packaged into a Docker image beforehand. The training environment includes the necessary operating system, data processing tools, monitoring and debugging tools, etc., which are not limited in this embodiment. After receiving the training instructions for the target model, the intelligent cockpit system automatically pulls the image and starts Docker, allowing the running of Linux binaries within the intelligent cockpit system to obtain the training environment required for the target model. A "one-click Docker" function can also be set up. Users can find and use the "one-click Docker" function in the application list of the intelligent cockpit system to send training instructions for the target model. The intelligent cockpit system responds to the training instructions for the target model by starting Docker to construct the training environment required for the target model.
[0042] In one specific embodiment, current smart cockpit systems generally use the Android system, so Docker can be adapted to the Android system based on the current smart cockpit system.
[0043] S102. In the training environment, train the target model according to the model training configuration parameters and model constraint configuration parameters.
[0044] The aforementioned model training configuration parameters include configuration information for the target model and its training data, specifically including the target model download path, target model storage path, training data download path, training data format requirements, and output path. It should be noted that users can modify the model training configuration parameters according to their actual needs. When using the intelligent cockpit system for model training for the first time, users can be guided to set the model training configuration parameters to facilitate training the target model according to these parameters.
[0045] The aforementioned model constraint configuration parameters include configuration information of available resources for the target model during training, specifically including computing power limitations, memory limitations, etc. In addition, it also includes displaying some basic conditions, such as the target model must be trained in the park position.
[0046] In some embodiments, the computing power limit and memory limit can be recommended by the manufacturer based on the actual situation, and users can also modify the computing power limit and memory limit according to their actual needs. This embodiment does not impose any limitations.
[0047] In the embodiments of this application, if the user needs to modify the model training configuration parameters and / or model constraint configuration, the parameters that need to be adjusted can be adjusted before starting training, and training can be started after the adjustment is completed. If the user does not need to modify the model training configuration parameters and model constraint configuration, training can be started directly.
[0048] In the training environment required by the target model, train the target model according to the model training configuration parameters and model constraint configuration parameters.
[0049] S103. During the training process, if the vehicle's operating status is detected to be inconsistent with the training requirements of the target model, a safety warning will be output.
[0050] It should be noted that since the target model is trained using the vehicle's intelligent cockpit system, it is necessary to ensure that model training does not affect the vehicle's normal operation. Specifically, the vehicle's operating status needs to be monitored during training. If the vehicle's operating status is detected to no longer meet the training requirements of the target model, a safety warning should be issued to remind the user to terminate the model training.
[0051] Generally, if a vehicle disengages from the parking gear, it indicates that the vehicle's operating state does not meet the training requirements of the target model. Specifically, if a vehicle disengages from the parking gear, it means the user needs to start the vehicle. To avoid the training model consuming excessive computing power and memory, which could affect the normal operation of the vehicle, a safety warning needs to be issued.
[0052] If the vehicle's battery level is too low, it indicates that the vehicle's operating status does not meet the training requirements of the target model. Specifically, if the vehicle's battery level is too low, energy consumption needs to be reduced to maintain vehicle operation, and a safety warning can also be issued. A battery level threshold can be set; if the vehicle's battery level is lower than this threshold, it indicates that the vehicle's operating status does not meet the training requirements of the target model. This battery level threshold can be set according to actual conditions, and this embodiment does not impose any limitations.
[0053] In some embodiments, if the user does not stop training the target model within a set time period after the safety warning is issued, the training of the target model can be automatically stopped and the user notified to ensure driving safety. This set time period can be set according to actual circumstances, and is not limited in this embodiment.
[0054] In the above embodiments, the intelligent cockpit system responds to the training command of the target model, constructs the training environment required by the target model, and trains the target model according to the model training configuration parameters and model constraint configuration parameters. The model training configuration parameters include the configuration information of the target model and its training data, and the model constraint configuration parameters include the configuration information of the available resources for the target model during training. During training, if the vehicle's operating state is detected to be inconsistent with the training requirements of the target model, a safety warning is output. This configuration enables the intelligent cockpit system to train the target model, thus meeting the computing power required for model training.
[0055] Furthermore, the intelligent cockpit system trains the target model according to the model training configuration parameters and model constraint configuration parameters, and outputs a safety warning when it detects that the vehicle's operating state does not meet the training requirements of the target model. This can effectively avoid the impact of training the target model on the normal operation of the vehicle and ensure the vehicle's driving safety.
[0056] As an optional implementation, the steps of the above embodiments respond to the training instructions of the target model and construct the training environment required by the target model, specifically including the following steps: In response to the training instructions of the target model, the training environment required for the target model is constructed using Docker or a virtual machine.
[0057] Specifically, in the embodiments of this application, in response to the training instructions of the target model, Docker or a virtual machine can be used to construct the training environment required for the target model in the intelligent cockpit system. The training environment required for model training includes the necessary operating system, data processing tools, monitoring and debugging tools, etc., which are not limited in this embodiment.
[0058] The specific process of constructing the training environment required for the target model using Docker can be referred to the description in the above embodiments, and will not be repeated here. Constructing the training environment required for the target model using a virtual machine can be achieved by installing the virtual machine as an application on the operating system of the intelligent cockpit system. After obtaining the training instructions for the target model, the virtual machine can be started to obtain the training environment required for the target model.
[0059] In the embodiments of this application, the training environment required for the target model is constructed in the vehicle's intelligent cockpit system using Docker or a virtual machine, which makes it possible to train the model in the intelligent cockpit system so as to make full use of the computing power of the intelligent cockpit system to meet the requirements of model training.
[0060] As an optional implementation, the method of the above embodiments further includes the following steps: During the training process of the target model, the training information of the target model is displayed; the training information includes at least one of the following: training time, training progress, current computing power used, and current memory used.
[0061] In the embodiments of this application, during the training process of the target model, the application interface of the intelligent cockpit system can also display the training information of the target model, including at least one of the following: training time, training progress, current computing power used, and current memory used.
[0062] This setup allows users to clearly understand the training progress of the target model, improving the user experience.
[0063] As an optional implementation, in the training process of the above embodiments, if it is detected that the vehicle's operating state does not meet the training requirements of the target model, a safety warning is output, specifically including the following steps: If the vehicle is detected to have exited the parking gear or the vehicle's battery level is detected to be lower than a first preset threshold, a first safety warning is issued; if the vehicle's battery level is detected to be lower than a second preset threshold, a second safety warning is issued and the training process of the target model is stopped; the first preset threshold is greater than the second preset threshold; the first safety warning is issued via push notification, and the second safety warning is issued via telephone reminder.
[0064] During training, the vehicle's operating status can be monitored, specifically its gear position and battery level.
[0065] If the system detects that the vehicle has exited the parking position or that the vehicle's battery level is below a first preset threshold, a first safety warning can be issued to ensure driving safety. If the system detects that the vehicle has exited the parking position, the first safety warning primarily informs the user that the target model needs to be trained in the parking position, and requests the user to switch back to the parking position or stop training the target model. If the system detects that the vehicle's battery level is below the first preset threshold, the first safety warning primarily informs the user that the current battery level is low, and requests the user to charge the battery or stop training the target model. When the battery level falls below a second preset threshold, training of the target model will automatically stop.
[0066] The first safety warning can be output via push notification. Specifically, after generating a first safety warning, the intelligent cockpit system can send it to the cloud server, which will then push the warning to the user's device via push notification. Alternatively, the intelligent cockpit system can also push the first safety warning directly to the user through the application interface of the intelligent cockpit system.
[0067] The first threshold value mentioned above can be set according to the actual situation, for example, it can be set to 20%, but this embodiment does not limit it.
[0068] If the vehicle's battery level is detected to be below a second preset threshold, a second safety warning is issued and the training process of the target model is stopped to ensure the vehicle can operate normally. The second safety warning is mainly used to inform the user that the current battery level is very low and that the training of the target model will be automatically stopped.
[0069] The second safety warning can be issued via telephone alert. After generating the second safety warning, the intelligent cockpit system can call the user through the vehicle management backend on the cloud server to inform the user that the battery is low, the training of the target model will be automatically stopped, and the user can be reminded to charge the battery as soon as possible.
[0070] The second set threshold is less than the first set threshold. The second set threshold can be set according to the actual situation, for example, it can be set to 5%. This embodiment does not limit this setting.
[0071] In the above embodiments, the operating status of the vehicle can be monitored to ensure that model training does not affect the normal operation of the vehicle.
[0072] Furthermore, after receiving the training instructions for the target model, the vehicle's operating status can be detected first. If the vehicle is not in the parking position or the vehicle's battery level is detected to be less than a second preset threshold, a third safety warning can be output to remind the user to place the vehicle in the parking position and ensure the battery level is greater than the second preset threshold before using the smart cockpit system to train the target model. This third safety warning can be forwarded to the user terminal via a cloud server or directly pushed to the smart cockpit system's display page; this embodiment does not impose any limitations on this.
[0073] In an optional embodiment, the model training configuration parameters of the above embodiments include the download path of the training data; the method of the above embodiments may further include the following steps: The training data can be downloaded from a storage device connected to the smart cockpit system. The storage device connects to the smart cockpit system via a network hotspot, the cloud, or a web interface that accesses the smart cockpit system.
[0074] The above model training configuration parameters include the download path for the training data.
[0075] Specifically, in the embodiments of this application, training data can be downloaded from an external storage device. That is, the download path for training data includes downloading training data from a storage device connected to the intelligent cockpit system.
[0076] A storage device refers to a hardware device used for long-term or temporary storage of data and information, capable of writing, reading, and storing data. The storage device can be a mobile phone, mobile computer, or USB flash drive, etc., and this embodiment is not limited to these. The storage device can be directly connected to the intelligent cockpit system via a data cable or USB interface. Alternatively, the storage device can also be connected to the intelligent cockpit system wirelessly.
[0077] Storage devices can connect to the smart cockpit system via a Wi-Fi hotspot. Specifically, users can activate the Wi-Fi hotspot on the smart cockpit system, and the storage device connects to the local area network (LAN) to establish a connection with the smart cockpit system. If an in-vehicle application (APP) is installed on the storage device, it can communicate with the cloud server, thereby establishing a connection with the smart cockpit system. Additionally, the smart cockpit system has a web interface, which the storage device can access to connect to the system.
[0078] Users can pre-establish a connection between the storage device and the smart cockpit system. When setting model training configuration parameters, they can specify the download path for training data to be downloaded from the storage device connected to the smart cockpit system. During target model training, the system can automatically retrieve training data from the storage device to train the target model, thereby reducing the memory usage of training data on the smart cockpit system and improving training speed.
[0079] In a specific embodiment, such as Figure 3 As shown, the training methods for the model include: Obtain the training instructions for the target model; Configure model training parameters and model limitation parameters; Train the target model; The system detects whether the vehicle is in the parking position. If the vehicle is not in the parking position, it sends a first safety warning to the user through the cloud server. If the vehicle is in the parking position, it checks whether the battery level is less than a second set threshold. If the battery level is less than the second set threshold, a second safety warning will be sent to the user through the vehicle management backend on the cloud server. If the battery level is not less than the second set threshold, the system will check whether the battery level is less than the first set threshold. If the battery level is less than the first set threshold, a first safety warning is sent to the user terminal via the cloud server; if the battery level is not less than the first set threshold, it is detected whether the target model has been trained. If the target model is trained successfully, training ends. If the target model is not trained successfully, the vehicle's operating status can be monitored repeatedly. Figure 3 As shown, this continues until the target model training is complete.
[0080] As an optional embodiment, the method of the above embodiments may further include the following steps: In response to a usage command from the in-vehicle local model, the in-vehicle local model is run; the usage command carries the target data that needs to be processed by the in-vehicle local model, and the sender of the usage command includes the user terminal that is connected to the intelligent cockpit system; the processing result of the in-vehicle local model on the target data is fed back to the user terminal.
[0081] The intelligent cockpit system in this embodiment can also provide computing power to run onboard local models.
[0082] The sender of the aforementioned in-vehicle local model usage instructions includes the user terminal, which is connected to the intelligent cockpit system. It is understood that the user terminal communicates with the intelligent cockpit system through a cloud server. The user can send usage instructions for the in-vehicle local model through the user terminal, and the cloud server receives these instructions and sends them to the intelligent cockpit system.
[0083] The instructions for using the vehicle-mounted local model carry the target data that needs to be processed by the vehicle-mounted local model. The target data can be photos, videos, text, or voice, etc., and this embodiment does not limit it.
[0084] In response to usage commands from the in-vehicle local model, the system runs the in-vehicle local model to process the target data. The intelligent cockpit system utilizes Docker or a virtual machine to construct a runtime environment for the in-vehicle local model, and runs the model within this environment. Target data can be input into the in-vehicle local model, and the system outputs the processed results, which are then fed back to the user. Specifically, the in-vehicle local model sends the processed results to the cloud server, which in turn sends the processed results to the user.
[0085] In one specific embodiment, a user can open the usage page of the in-vehicle local model on the user's mobile phone manufacturer's app, select the name of the in-vehicle local model to be used and enter the target data, and then click OK or Run to generate the usage instructions for the in-vehicle local model and send the usage instructions for the in-vehicle local model to the cloud server.
[0086] The cloud server sends usage instructions for the in-vehicle local model to the intelligent cockpit system. The intelligent cockpit system responds to these instructions by running the in-vehicle local model, processing the target data, and obtaining the processing results. The intelligent cockpit system then sends the processing results back to the cloud server, which in turn sends the results back to the user's device.
[0087] This setup allows some models that cannot run or run slowly on the user's device to run on the smart cockpit system, providing convenience for users.
[0088] In one specific embodiment, such as Figure 4 As shown, users can select the desired in-vehicle local model through the user terminal, then select the target data, click "Run" to generate usage instructions for the in-vehicle local model carrying the target data, and then send the usage instructions for the in-vehicle local model to the cloud server.
[0089] The cloud server forwards the usage instructions of the onboard local model to the intelligent cockpit system.
[0090] The intelligent cockpit system utilizes Docker or a virtual machine to construct a runtime environment for an onboard local model, runs the onboard local model, processes the target data, and obtains the processing results. These results are then sent to a cloud server.
[0091] The cloud server forwards the processing results of the target data to the user terminal.
[0092] Corresponding to the aforementioned application function implementation method embodiments, this application also provides a model training device, electronic device, computer-readable storage medium, computer program product, and corresponding embodiments.
[0093] Figure 5 This is a schematic diagram of the structure of the training device for the model shown in the embodiments of this application.
[0094] See Figure 5 The training apparatus for the model includes: Construction module 100 is used to construct the training environment required by the target model in response to the training instructions of the target model; The training module 110 is used to train the target model in the training environment according to the model training configuration parameters and the model constraint configuration parameters. The model training configuration parameters include the configuration information of the target model and the training data of the target model, and the model constraint configuration parameters include the configuration information of the resources available to the target model during the training process. The early warning module 120 is used to output a safety warning if the vehicle's operating status does not meet the training requirements of the target model during the training process.
[0095] Furthermore, the construction module 100 in the above embodiment, when constructing the training environment required by the target model in response to the training instruction of the target model, is specifically used for: In response to the training instructions of the target model, the training environment required for the target model is constructed using Docker or a virtual machine.
[0096] Furthermore, the training apparatus for the model in the above embodiments further includes: The display module is used to display the training information of the target model during the training process; the training information includes at least one of the following: training time, training progress, current computing power used, and current memory used.
[0097] Furthermore, the warning module 120 in the above embodiment, when outputting a safety warning if it detects that the vehicle's operating state does not meet the training requirements of the target model, is specifically used for: If the vehicle is detected to have exited the parking gear and / or the vehicle's battery level is detected to be less than a first preset threshold, a first safety warning is issued; if the vehicle's battery level is detected to be less than a second preset threshold, a second safety warning is issued and the training process of the target model is stopped; the first preset threshold is greater than the second preset threshold; the first safety warning is issued via push notification, and the second safety warning is issued via telephone reminder.
[0098] Furthermore, the training apparatus for the model in the above embodiments further includes: The operation module is used to respond to the usage instructions of the in-vehicle local model, run the in-vehicle local model; the usage instructions carry the target data that needs to be processed by the in-vehicle local model, and the sender of the usage instructions includes the user terminal that is connected to the intelligent cockpit system; and the processing results of the in-vehicle local model on the target data are fed back to the user terminal.
[0099] Furthermore, the model training configuration parameters in the above embodiments include the download path of the training data; the download path of the training data includes downloading the training data from a storage device connected to the intelligent cockpit system; the storage device establishes a connection with the intelligent cockpit system through a network hotspot, the cloud, or by accessing the web interface of the intelligent cockpit system.
[0100] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated further here.
[0101] Figure 6 This is a schematic diagram of the structure of an electronic device shown in an embodiment of this application.
[0102] See Figure 6 The electronic device includes a memory 200 and a processor 210.
[0103] The electronic device includes a memory 200 and a processor 210; The memory 200 is connected to the processor 210 and is used to store programs; Processor 210 is used to implement some or all of the methods described above by running programs stored in memory 200.
[0104] Specifically, the aforementioned electronic device may further include: a bus, a communication interface 220, an input device 230, and an output device 240.
[0105] The processor 210, memory 200, communication interface 220, input device 230, and output device 240 are interconnected via a bus. Among them: A bus can include a pathway for transmitting information between various components of a computer system.
[0106] The processor 210 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.
[0107] Processor 210 may include a main processor, as well as a baseband chip, modem, etc.
[0108] Memory 200 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM may store static data or instructions required by processor 210 or other modules of the computer. Permanent storage devices may be read-write storage devices. Permanent storage devices may be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices employ mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices may be removable storage devices (e.g., floppy disks, optical drives). System memory may be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory may store some or all of the instructions and data required by the processor during operation. Furthermore, memory 200 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, memory 200 may include a removable storage device that is readable and / or writable, such as a laser disc (CD), a read-only digital multifunction optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, an ultra-high density optical disc, a flash memory card (e.g., SD card, mini SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not contain carrier waves or transient electronic signals transmitted wirelessly or via wired connections.
[0109] Input device 230 may include a device for receiving user input data and information, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.
[0110] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.
[0111] The communication interface 220 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.
[0112] The processor 210 executes the program stored in the memory 200 and calls other devices, and can be used to implement some or all of the methods described above.
[0113] Another embodiment of this application also proposes a vehicle, see [link to relevant documentation] Figure 7 As shown, the vehicle includes an intelligent cockpit system 300, which is configured to implement some or all of the methods described above.
[0114] The vehicle provided in this embodiment belongs to the same application concept as the model training method provided in the above embodiments of this application. It can execute the model training method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the above model training method. For technical details not described in detail in this embodiment, please refer to the specific processing content of the model training method provided in the above embodiments of this application, which will not be repeated here.
[0115] Furthermore, the method according to this application can also be implemented as a computer program product, which includes computer program code instructions for performing some or all of the steps in the method described above. Optionally, the computer program can be stored on a readable storage medium of a computer device or in the cloud; the processor of the computer device reads the computer program from the readable storage medium or the cloud.
[0116] Computer program products can be written in any combination of one or more programming languages to perform the operations of the embodiments of this application. The programming languages include object-oriented programming languages such as Java and C++, as well as conventional procedural programming languages such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.
[0117] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.
[0118] Alternatively, this application may be implemented as a computer-readable storage medium (or a non-transitory machine-readable storage medium or a machine-readable storage medium) storing executable code (or computer program or computer instruction code) that, when executed by a processor of an electronic device (or server, etc.), causes the processor to perform part or all of the steps of the methods described above according to this application.
[0119] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0120] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.
Claims
1. A method for training a model, characterized in that, A smart cockpit system applied to vehicles, the method comprising: In response to the training instructions of the target model, construct the training environment required for the target model; In the training environment, the target model is trained according to the model training configuration parameters and the model constraint configuration parameters; the model training configuration parameters include the configuration information of the target model and the training data of the target model, and the model constraint configuration parameters include the configuration information of the resources available to the target model during the training process; During training, if the vehicle's operating status is detected to be inconsistent with the training requirements of the target model, a safety warning is output.
2. The method according to claim 1, characterized in that, The step of constructing the training environment required for the target model in response to training instructions includes: In response to the training instructions for the target model, the training environment required for the target model is constructed using Docker or a virtual machine.
3. The method according to claim 1, characterized in that, Also includes: During the training process of the target model, the training information of the target model is displayed; The training information includes at least one of the following: training time, training progress, current computing power used, and current memory used.
4. The method according to claim 1, characterized in that, During the training process, if the vehicle's operating state is detected to be inconsistent with the training requirements of the target model, a safety warning is output, including: If the vehicle is detected to have exited the parking position or the vehicle's battery level is detected to be less than a first preset threshold, a first safety warning is issued; if the vehicle's battery level is detected to be less than a second preset threshold, a second safety warning is issued and the training process of the target model is stopped; the first preset threshold is greater than the second preset threshold. The first security alert is output via message push, and the second security alert is output via telephone reminder.
5. The method according to claim 1, characterized in that, Also includes: In response to a usage instruction from the in-vehicle local model, the in-vehicle local model is run; the usage instruction carries target data that needs to be processed by the in-vehicle local model, and the sender of the usage instruction includes a user terminal that is communicatively connected to the intelligent cockpit system. The processing results of the target data by the in-vehicle local model are fed back to the user terminal.
6. The method according to claim 1, characterized in that, The model training configuration parameters include the download path of the training data; The download path for the training data includes downloading the training data from a storage device connected to the intelligent cockpit system; the storage device establishes a connection with the intelligent cockpit system via a network hotspot, the cloud, or by accessing the web interface of the intelligent cockpit system.
7. A training device for a model, characterized in that, A smart cockpit system for vehicles, the device comprising: A construction module is used to construct the training environment required by the target model in response to the training instructions of the target model; A training module is used to train the target model in the training environment according to model training configuration parameters and model constraint configuration parameters; the model training configuration parameters include configuration information of the target model and the training data of the target model, and the model constraint configuration parameters include configuration information of the resources available to the target model during training; The early warning module is used to output a safety warning if the vehicle's operating status does not meet the training requirements of the target model during the training process.
8. An electronic device, characterized in that, include: processor; as well as A memory having executable code stored thereon, which, when executed by the processor, causes the processor to perform the method as described in any one of claims 1-6.
9. A vehicle, characterized in that, include: Intelligent cockpit system; The intelligent cockpit system is configured to implement the method of any one of claims 1-6.
10. A computer program product, characterized in that, The computer program product includes computer instructions that, when executed by a processor, implement the method described in any one of claims 1-6.