TW202133054A - Transformation method of neural network model, model accuracy positioning method and apparatuses - Google Patents

Abstract

The present disclosure provides a transformation method of a neural network model, a model accuracy positioning method and apparatuses. In the transformation scheme of the neural network model, a model transformation apparatus may automatically obtain an accuracy verification result of a second neural network model based on an inference result obtained by a first neural network model under a source frame for model input data and an inference result obtained by a second neural network model under a target frame for the model input data, thereby improving the accuracy verification efficiency of the transformation of the neural network model. In the scheme of model accuracy positioning, the model accuracy positioning apparatus may locate a target network layer in the second neural network model with unaligned accuracy based on the output data of each network layer obtained by the first neural network model under the source frame and the second neural network model under the target frame for the same model input data, respectively, thereby achieving accurate positioning of the network layer with unaligned accuracy.

Description

Neural network model conversion method, model accuracy positioning method and device

The present invention relates to the technical field of artificial intelligence (AI), in particular to a neural network model conversion method, model accuracy positioning method and device.

At present, the industry has developed a variety of deep learning model frameworks, each with its own network structure definition and toolkit. Since different frameworks have their own advantages in training, deployment, platform compatibility, etc., it is often necessary to convert deep learning models between frameworks. The conversion of a neural network model refers to the conversion of a neural network model defined in a certain framework to another framework.

However, after the neural network model is converted, it is necessary to verify the accuracy of the model manually by the programmer, and the accuracy of the verification is inefficient.

The invention provides a technical solution for neural network model conversion.

In a first aspect, a neural network model conversion method is provided. The method includes: a model conversion device obtains a first neural network model in a source framework, model input data, and information about a target framework; the model conversion device is based on the Model input data performs model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework; the model conversion device is based on the first neural network model and the The second neural network model separately processes the first inference result and the second inference result obtained by processing the model input data to obtain the accuracy verification result of the second neural network model.

The model conversion device may perform model conversion processing based on a user instruction. The user instruction is used to instruct to convert the first neural network model in the source frame to the target frame. The user instruction includes the first neural network in the source frame. Road model, model input data, and target framework information. The model conversion device can also perform model conversion processing based on internal operating instructions, such as model conversion instructions. The internal operating instructions are used to instruct to convert the first neural network model under the source frame to the target frame. The internal operating instructions Including the first neural network model under the source frame, model input data, and information about the target frame.

In a possible implementation, the model conversion device obtains the first inference result of the first neural network model on the model input data in the process of performing model conversion processing on the first neural network model.

In a possible implementation, the model conversion device stores the first inference result in the local storage area.

In a possible implementation, the model conversion device processes the model input data through the second neural network model to obtain a second reasoning result.

In a possible implementation, the model conversion device obtains the accuracy verification result of the second neural network model by comparing the first inference result with the second inference result.

In a possible implementation, the performing model conversion processing on the first neural network model under the source framework to obtain the second neural network model under the target framework includes: performing a first process on the The first neural network model in the source framework performs model conversion processing to obtain the second neural network model in the target framework; the method further includes: creating a second process through the first process; A process calls the second process to process the model input data through the second neural network model in the second process to obtain a second inference result.

In a possible implementation, the model conversion device is based on a first inference result and a second inference result obtained by separately processing the model input data on the first neural network model and the second neural network model , Obtaining the accuracy verification result of the second neural network model, including: if the difference between the first inference result and the second inference result is within a preset error range, determining the second neural network The accuracy verification result of the model is accuracy alignment; or if the difference between the first inference result and the second inference result exceeds the preset error range, the accuracy verification result of the second neural network model is determined to be The precision is not aligned.

In a possible implementation, the method further includes: in response to determining that the accuracy verification result of the second neural network model is accuracy misalignment, the model conversion device locates the accuracy in the second neural network model. Aligned target network layer; the model conversion device outputs information of the target network layer.

In a possible implementation, the locating a target network layer whose accuracy is not aligned in the second neural network model includes: establishing the first neural network model and the second neural network by the model conversion device The mapping relationship between the network layers in the network model; each network in the process of separately processing the input data of the model based on the first neural network model and the second neural network model by the model conversion device The output data of the path layer and the mapping relationship determine the target network layer whose accuracy is not aligned in the second neural network model.

In a possible implementation, the target network layer is the network layer in which the accuracy is not aligned first among the multiple network layers included in the second neural network model.

In another possible implementation, the target network layer is a plurality of network layers whose accuracy is not aligned in the second neural network model. For example, it includes not only the network layer where the accuracy misalignment occurs first, but also the subsequent network layers of the network layer.

In a possible implementation, the information of the target network layer includes at least one of the following: the identification, parameters, output data of the target network layer, the output data and the first neural network model The error between the output data of the network layer corresponding to the target network layer.

In a possible implementation, the model conversion device performs model conversion processing on the first neural network model again, wherein, in the process of performing the model conversion processing again, the first neural network model is acquired Output data of each network layer in and/or establish a mapping relationship between the network layers in the first neural network model and the second neural network model.

In a possible implementation, the model conversion device performs model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework, including: the model conversion The device creates a network topology map by processing the input data of the model by the first neural network model, wherein the network topology map includes the information of the network layer in the first neural network model and The connection relationship of the network layer; the model conversion device obtains the second neural network model based on the network topology map and the information of the target frame.

In a second aspect, a model conversion device is provided. The device includes: an acquiring unit for acquiring a first neural network model, model input data, and information about a target frame under a source frame; Perform model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework; the verification unit is configured to be based on the first neural network model and the second neural network The first inference result and the second inference result obtained by processing the input data of the model respectively by the path model obtain the accuracy verification result of the second neural network model.

In a third aspect, an electronic device is provided, including a memory and a processor, wherein the memory stores program instructions, and the processor is used to call the program instructions stored in the memory to execute the first aspect or Any one of the first aspect implements the method described.

In a fourth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions that, when run on a computer, cause the computer to execute the first aspect or any one of the first aspects described above. Methods.

In a fifth aspect, a computer program product containing instructions is provided, which when run on a computer, causes the computer to execute the method described in the first aspect or any one of the first aspects.

The conversion scheme using the neural network model of the present invention has the following technical effects:

According to the inference results of the first neural network model under the source framework and the second neural network model under the target framework on the same model input data, the accuracy verification results of the second neural network model can be automatically obtained, which improves the neural network. The accuracy of network model conversion verifies the efficiency.

The embodiment of the present application also provides a model accuracy positioning solution.

In a sixth aspect, a model accuracy positioning method is provided, including: a model accuracy positioning device acquires a first neural network model, model input data, and information about a target frame under a source frame; The neural network model performs model conversion processing to obtain a second neural network model under the target framework; the model accuracy positioning device performs a model conversion process on the neural network model based on the first neural network model and the second neural network model The output data of each network layer obtained during the processing of the model input data is used to locate the target network layer whose accuracy is not aligned in the second neural network model; the model accuracy positioning device outputs the target network Layer of information.

In the case where the accuracy of the second neural network model is known to be misaligned with the first neural network model, the model accuracy positioning device can perform model accuracy positioning based on a user instruction, which is used to instruct the positioning of the second neural network The target network layer in the road model with misaligned accuracy, the user command includes the first neural network model under the source frame, model input data, and information about the target frame. For example, when the user obtains the result that the accuracy of the second neural network model and the first neural network model are not aligned, the user instruction is sent to the model accuracy positioning device. The model accuracy positioning device can also perform model accuracy positioning based on internal operating instructions, such as model accuracy positioning instructions. For example, when the model accuracy positioning device obtains an instruction that the accuracy of the second neural network model and the first neural network model are not aligned, it triggers the model accuracy positioning instruction. The internal operation instructions are used to instruct to locate the target network layer of the second neural network model whose accuracy is not aligned. The internal operation instructions include the first neural network model under the source frame, model input data, and information about the target frame . In a possible implementation manner, the input data of the model is processed through the second neural network model to obtain the output data of each network layer in the second neural network model.

In a possible implementation manner, the output data of each network layer where the first neural network model processes the model input data is obtained in the process of the model conversion processing.

In a possible implementation, the target network layer is the network layer in which the accuracy is not aligned first among the multiple network layers included in the second neural network model.

In another possible implementation, the target network layer is a plurality of network layers whose accuracy is not aligned in the second neural network model. For example, it includes not only the network layer where the accuracy misalignment occurs first, but also the subsequent network layers of the network layer.

In a possible implementation, the target network layer can be determined by comparing the output data of each network layer in the first neural network model and the second neural network model.

In a possible implementation, if the difference between the output data of the network layer included in the second neural network model and the corresponding network layer of the first neural network model exceeds a preset error range, then The network layer is considered to be a network layer with misaligned accuracy.

In a possible implementation, the information of the target network layer includes at least one of the following: the identification, parameters, output data of the target network layer, the output data and the first neural network model The error between the output data of the network layer corresponding to the target network layer.

In a possible implementation, the method further includes: the model accuracy positioning device establishes one of the network layers in the first neural network model and the second neural network model based on the model conversion processing The model accuracy positioning device is based on the output data of each network layer obtained by processing the input data of the model based on the first neural network model and the second neural network model, and locates the output data of each network layer The target network layer whose accuracy is not aligned in the second neural network model includes: the model accuracy locating device is based on each network layer in the first neural network model and the second neural network model The output data and the mapping relationship determine the target network layer whose accuracy is not aligned in the second neural network model.

In a possible implementation, the model accuracy positioning device determines the first neural network model and the second neural network model based on the output data of each network layer and the mapping relationship. 2. Target network layers with misaligned accuracy in the neural network model, including: the model accuracy positioning device sequentially compares corresponding networks in the first neural network model and the second neural network model based on the mapping relationship Output data of the circuit layer; in response to the difference between the output data of the corresponding network layer in the first neural network model and the second neural network model exceeding a preset error range, the corresponding network layer The network layer in the second neural network model is determined as the target network layer.

In a possible implementation, the model accuracy positioning device performs model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework, including: the model The precision positioning device creates a network topology map by processing the input data of the model by the first neural network model, wherein the network topology map includes the network layer in the first neural network model The connection relationship between the information and the network layer; the model accuracy positioning device obtains the second neural network model based on the network topology map and the information of the target frame.

In a seventh aspect, a model accuracy positioning device is provided, including: an acquiring unit for acquiring a first neural network model, model input data, and information about a target frame under a source frame; a conversion unit for comparing the first neural network model The neural network model performs model conversion processing to obtain a second neural network model under the target framework; a positioning unit is used to process the neural network model based on the first neural network model and the second neural network model respectively The output data of each network layer in the process of model input data locates the target network layer whose accuracy is not aligned in the second neural network model; the output unit is used to output the information of the target network layer.

In an eighth aspect, there is provided an electronic device including a memory and a processor, wherein the memory stores program instructions, and the processor is used to call the program instructions stored in the memory to execute the sixth aspect or the first aspect. Any one of the six aspects implements the described method.

In a ninth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions that, when run on a computer, cause the computer to execute any one of the sixth aspect or the sixth aspect described above. Methods.

In a tenth aspect, a computer program product containing instructions is provided, which when run on a computer, causes the computer to execute the method described in any one of the sixth aspect or the sixth aspect.

Adopting the model accuracy positioning scheme of the present invention has the following technical effects:

According to the output data of each network layer obtained by the first neural network model under the source framework and the second neural network model under the target framework on the input data of the same model, the accuracy of the second neural network model can be located. The aligned target network layer realizes the accurate positioning of the network layer when the accuracy is not aligned.

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

It should be understood that the term "and/or" in the present invention is only an association relationship describing related objects, which means that three relationships can exist. For example, A and/or B can mean that A alone exists, and both A and B exist. There are three cases of B alone. In addition, the character "/" in this article generally indicates that the related objects before and after are in an "or" relationship. The terms "first", "second", etc. in the present invention are used to distinguish different objects, rather than to describe a specific order.

Please refer to FIG. 1, which is a schematic flowchart of a neural network model conversion method provided by an embodiment of the present invention. The method includes the following steps:

S101. The model conversion device obtains the first neural network model under the source frame, the model input data, and the information of the target frame.

Because different frameworks have their own advantages in training, deployment, platform compatibility, etc., neural network models can be converted between frameworks. Each framework has its own network structure definition and toolkit. The conversion of a neural network model refers to the conversion of a neural network model defined in a certain framework to another framework, for example, the neural network model under the tensorflow framework is converted to the pytorch framework, or the custom framework The following neural network model is converted to a more general framework such as pytorch or tensorflow, or vice versa, and so on. The corresponding frame before the neural network model conversion is called the "source frame", and the corresponding frame after the neural network model conversion is called the "target frame". The embodiment of the present invention does not limit the specific implementation of the source frame and the target frame.

The model conversion device may perform model conversion processing based on a user instruction. The user instruction is used to instruct to convert the first neural network model in the source frame to the target frame. The user instruction includes the first neural network in the source frame. Road model, model input data, and target framework information. The model conversion device can also perform model conversion processing based on internal operating instructions, such as model conversion instructions. The internal operating instructions are used to instruct to convert the first neural network model under the source frame to the target frame. The internal operating instructions Including the first neural network model under the source frame, model input data, and information about the target frame. Among them, the first neural network model is a general neural network model, which can be a neural network model with any function. The first neural network model includes one or more network layers, such as a convolutional layer, a pooling layer, a fully connected layer, a normalization layer, and so on. The first neural network model under the source framework includes the structure description file of each network layer in the first neural network model under the source framework, etc. The structure description file is used to describe the name of the network layer and the layer Index etc. The model input data refers to the model input data used for neural network model inference, or can be called test data. The input data of the model can be a picture, a paragraph of text, etc. The target framework information includes the target framework's logo, format, toolkit, etc.

S102. The model conversion device performs model conversion processing on the first neural network model in the source frame to obtain the second neural network model in the target frame.

To obtain the second neural network model under the target frame, the first neural network model under the source frame needs to be subjected to model conversion processing, that is, the first neural network model is converted to the target frame. As shown in FIG. 2, it is a schematic diagram of the conversion process of the neural network model provided by the embodiment of the present invention. A model conversion device (also called a neural network model conversion tool) can be used to perform the frame conversion of the neural network model. Use the neural network model conversion tool to convert the network layers of the first neural network model under the source framework into the format of the target framework, and obtain the network layers of the second neural network model under the target framework, thereby obtaining The second neural network model under the target framework, where one network layer of the first neural network model may be converted into one or more network layers of the second neural network model, and the second neural network model One network layer may correspond to one or more network layers of the first neural network model, which is not limited in the embodiment of the present invention.

In some embodiments, in the process of performing the above-mentioned model conversion processing on the first neural network model, the model conversion device runs the inference process of the first neural network model on the input data of the model to obtain the first inference result, such as a graph 2 shows the output 1. Optionally, the inference result can be stored or recorded locally as the first inference result. The first inference result can be used to verify the accuracy of the second neural network model. That is, in the process of model conversion, the inference result of the first neural network model on the input data of the model can be obtained without re-running the inference process. Thereby improving the efficiency of accuracy verification.

In other embodiments, the model conversion device may also re-input the model input data to the first neural network model, and run the inference process of the first neural network model on the model input data to obtain the first inference result.

In other embodiments, the model conversion device may pre-store the first inference result of the first neural network model on the model input data.

The model conversion device performs model conversion processing on the first neural network model in the source frame to obtain the second neural network model in the target frame. Specifically, first, the model conversion device performs model analysis processing on the first neural network model to obtain a network topology diagram of the first neural network model. Optionally, the model analysis processing can be performed through the first neural network model. The model is obtained by reasoning on the input data of the model or obtained by other methods. The network topology diagram can also be called a computational diagram, which can include each network layer in the first neural network model and the connection relationship between each network layer, for example, as shown in the figure The network topology shown in 4, wherein the network topology diagram may not include the output of each network layer, which is beneficial to improve the model conversion efficiency.

Then, the model conversion device obtains the second neural network model based on the network topology map and the information of the target frame. After the model conversion device obtains the network topology diagram of the first neural network, the target framework has its own network structure definition and toolkit. The model conversion device can be based on the network topology diagram of the first neural network and use the network topology diagram of the target framework. The network structure definition and/or toolkit performs adaptive network construction on the network topology map of the first neural network to obtain the second neural network model.

S103. The model conversion device obtains a first inference result and a second inference result obtained by separately processing the model input data based on the first neural network model and the second neural network model to obtain the accuracy verification result of the second neural network model.

After obtaining the second neural network model under the target framework, the above model input data can be input to the second neural network model, and the model input data can be used to run on each network layer of the second neural network model Inferring again, the inference result output by the previous network layer can be used as the input data of the next network layer, and finally the final inference result of the second neural network model is obtained, that is, the second inference result.

It is understandable that the first inference result is the inference result of the output of the last layer obtained by processing the input data of the model by the first neural network model in the source framework or the meaningful inference result of a certain reciprocal layer, correspondingly , The second inference result is the inference result of the last layer of output obtained by processing the input data of the model by the second neural network model under the target framework or the meaningful inference result of a certain reciprocal layer, which is the embodiment of the present invention Not limited.

As shown in Figure 2, the model input data is input into the first neural network model under the source frame and the second neural network model under the target frame, respectively, and output 1 (that is, the above-mentioned first inference result) and output 2 (that is, the above-mentioned first inference result) are obtained respectively The second reasoning result above).

After obtaining the first inference result and the second inference result, because the first inference result and the second inference result are obtained by using the same model input data to run the inference in the first neural network model and the second neural network model, respectively Therefore, according to the first inference result and the second inference result, for example, by comparing the first inference result and the second inference result, the accuracy verification result of the second neural network model under the target framework can be obtained. For another example, multiple parameter values of the second inference result can be matched with the corresponding parameter values in the first inference result one by one to obtain the matching result, that is, the aforementioned accuracy verification result can be obtained. In some embodiments, the accuracy verification result may indicate a comparison result of the model accuracy of the second neural network model and the first neural network model, such as accuracy alignment or accuracy misalignment, or the accuracy verification result may indicate the second neural network model. Whether the model accuracy of the network model meets expectations or a set standard, for example, the accuracy meets the standard or the accuracy does not meet the standard, etc., the embodiment of the present invention does not limit this.

In some embodiments, the accuracy verification result may be output, or the information of the second neural network model may also be saved, and the embodiment of the present invention is not limited to this.

As shown in Figure 2, the accuracy verification is performed based on the output 1 of the neural network model under the source framework and the output 2 of the neural network model under the target framework.

According to a neural network model conversion method provided by an embodiment of the present invention, according to the inference results obtained by the first neural network model in the source frame and the second neural network model in the target frame on the input data of the same model, The accuracy verification result of the second neural network model is automatically obtained, which improves the accuracy verification efficiency of the neural network model conversion.

Please refer to FIG. 3, which is a schematic flowchart of another method for transforming a neural network model provided by an embodiment of the present invention. The method includes the following steps:

S201: The model conversion device obtains the first neural network model under the source frame, the model input data, and the information of the target frame.

For the specific implementation of this step, refer to step S101 in the embodiment shown in FIG. 1.

S202: Perform model conversion processing on the first neural network model under the source framework through the first process to obtain the second neural network model under the target framework, and record the first neural network model obtained by processing the input data of the model by the first neural network model. One reasoning result.

Part of the implementation of this step can refer to step S102 in the embodiment shown in FIG. 1.

The above-mentioned model conversion process is carried out in the first process.

S203: Create a second process through the first process.

When using the same model input data to run inference in the second neural network model, in order to prevent resource conflicts during the running of the inference process, a new process can be created through the first process, that is, the second process. The resources occupied by a process are different.

It is understandable that in multi-process calls, the neural network model conversion tool and the first neural network model under the source framework are generally developed based on a certain platform, but when the neural network model is converted, Different execution methods will be selected according to different target frameworks. Therefore, the second process created may be a process under the system environment or a process under the platform. In some embodiments, the neural network model conversion tool and the source framework can be developed based on the Python platform. When the target framework is the Caffe framework, since Caffe provides the Python interface of pycaffe, you can use the Python multiprocessing module to create a new process to continue in Use the target framework in the Python environment. In some embodiments, the neural network model conversion tool and the source framework can be developed based on the Python platform. When the target framework supports the invocation of the command line, the os.system of the Python os module can be used to create a new process in the system environment. Execute the command to start the target framework.

By creating a new process and running the second neural network model in the new process to reason about the input data of the model, you can avoid the second neural network model and the first neural network model in the same process during the inference process. The conflict of certain system calls caused the program to crash. Improve the reliability of the reasoning process.

S204. Invoke the second process through the first process, so that the second process executes the processing of the second neural network model on the input data of the model to obtain a second inference result.

Invoke the newly created second process through the first process, input the above model input data into the second neural network model, and use the model input data to run inferences on each network layer on the second neural network model. The inference result output by one network layer can be used as the input data of the next network layer, and finally the final inference result of the second neural network model is obtained, that is, the second inference result.

S205: Determine whether the difference between the first inference result and the second inference result is within a preset error range, if so, proceed to step S206; if not, proceed to S207.

S206: Determine that the accuracy verification result of the second neural network model is accuracy alignment.

S207: Determine that the accuracy verification result of the second neural network model is accuracy misalignment.

After obtaining the first inference result and the second inference result, because the first inference result and the second inference result are obtained by using the same model input data to run the inference in the first neural network model and the second neural network model, respectively Therefore, according to the first inference result and the second inference result, the accuracy verification result of the second neural network model under the target framework can be obtained. Specifically, if the difference between the first inference result and the second inference result is within a preset error range, it is determined that the accuracy verification result of the second neural network model is accuracy alignment. If the difference between the first inference result and the second inference result exceeds the preset error range, it is determined that the accuracy verification result of the second neural network model is accuracy misalignment.

S208. In response to determining that the accuracy verification result of the second neural network model is accuracy misalignment, the model conversion device locates the target network layer of the second neural network model whose accuracy is not aligned.

When it is determined that the accuracy verification result of the second neural network model is accuracy misalignment, it is necessary to locate the network layer in the first neural network model and the second neural network model where the accuracy is not aligned first, that is, the positioning first appears the reasoning The network layer with inconsistent results, the located network layer is called the target network layer. Wherein, the target network layer is the network layer in which the accuracy is not aligned first among the multiple network layers included in the second neural network model. In other words, the target network layer is the first one of the multiple network layers included in the second neural network model where the difference between the output data of the corresponding network layer of the first neural network model exceeds the preset error range Network layer.

In order to improve the efficiency of the neural network model conversion tool, in the accuracy verification process of the aforementioned neural network model, the inference results (also called output data) of any layer are generally not saved. When the accuracy verification result is that the accuracy is not aligned, the neural network model conversion tool performs a neural network model conversion and inference again to locate the accuracy problem. That is, the model conversion device performs model conversion processing on the first neural network model again. The output data of each layer of the neural network model is relatively large. Generally, the output data of each layer of the neural network model is not stored in the memory, but the output data is localized, that is, stored locally. Therefore, in the process of performing the model conversion process again, the output data of each network layer in the first neural network model is saved locally. When verifying the accuracy of a certain network layer, load the output data of the network layer from the local into the memory.

Since the neural network model conversion tool is used to convert each network layer of the first neural network model under the source framework into the format of the target framework, the source framework will use the model to input data in the first neural network model. Run the reasoning on it once to obtain the information of the network layer in the first neural network model and the connection relationship of the network layer. Then, the model conversion device can use the network structure definition of the target framework and the toolkit to adaptively configure the network topology of the first neural network model based on the network topology of the first neural network model to obtain the first neural network model. 2. Neural network model. Therefore, further, in the process of performing the model conversion process again, the mapping relationship between the network layers in the first neural network model and the second neural network model can also be established. Among them, the mapping relationship between the network layers of the first neural network model and the second neural network model can be one-to-many, many-to-one, or one-to-one. Accordingly, the target network layer can be one or more Network layer.

Then input the same model input data into the second neural network model, and record the output data of each network layer in the process of processing the model input data by the second neural network model.

The model conversion device obtains the output data of each network layer in the first neural network model and the second neural network model, and the difference between the network layers in the first neural network model and the second neural network model After the mapping relationship between the two, based on the output data and mapping relationship of each network layer in the first neural network model and the second neural network model, it can be determined that the target network with misaligned accuracy first appears in the second neural network model. Road layer.

Specifically, in the accuracy positioning process, in order to accurately locate the target network layer whose accuracy is not aligned first, a serial number is assigned to each network layer in the first neural network model. The sequence number can be in ascending powers, so that when positioning accuracy is a problem, each network layer in the first neural network model can be called in turn according to the sequence numbers of multiple network layers in the first neural network model.

For each network layer called, the inference result output by the previous network layer is input to the next network layer for inference, and the inference result of each network layer is obtained. Moreover, according to the serial number of each network layer, the inference result of each network layer is correspondingly stored locally. And also identify the inference results of each network layer. Among them, a network layer may have multiple inputs. Only save the inference results of each layer output, because in addition to the initial input data, the inputs of other layers are the output of the previous layer, so once the initial input of the first neural network model and the output of each layer are recorded, it is equivalent to recording The input and output of all layers are included. As shown in FIG. 4, it is a schematic diagram of the network topology obtained in the process of locating the target network layer according to the embodiment of the present invention. The neural network model includes multiple network layers: convolutional layer, activation function, normalization layer, Eltwise layer and pooling layer, etc. After the initial input data to the first network layer of the neural network model, the output of each network layer is used as the input of the next network layer. The inference results of the neural network model include initial input data and input/output data 1-10.

Each network layer in the first neural network model is individually converted to the target framework to obtain the second neural network model. Input the same model input data to the second neural network model, and then use the input of each layer to run the inference results of each layer of the model in the second neural network model, and save it locally. The serial number is the index. Of course, the reasoning under the target framework also needs to be carried out in a new process.

After obtaining the inference results of each network layer of the neural network model in the source frame and the target frame, you can compare each layer in the first neural network model in the source frame according to the sequence number of each network layer. The inference result of a network layer and the inference result of the corresponding network layer in the second neural network model under the target framework are located, and the network in the second neural network model whose accuracy is not aligned with the first neural network model first Road layer, that is, the target network layer.

S209. The model conversion device outputs the information of the target network layer.

After determining the target network layer, the model conversion device can output the information of the target network layer. The information of the target network layer includes the identification, parameters, input data, output results, etc. of the network layer. The administrator or user can analyze the reason for the misalignment of accuracy based on the information of the target network layer.

According to a neural network model conversion method provided by an embodiment of the present invention, according to the inference results obtained by the first neural network model under the source frame and the second neural network model under the target frame on the same model input data, it is possible to The accuracy verification result of the second neural network model is automatically obtained, which improves the accuracy verification efficiency of the neural network model conversion; and according to the first neural network model under the source frame and the second neural network model under the target frame The output data of each network layer obtained from the model input data can locate the target network layer whose accuracy is not aligned in the second neural network model, and realize the accurate positioning of the network layer when the accuracy is not aligned.

When converting the neural network model of the source frame to the target frame, input data to the same model of the neural network model, and judge whether the deviation between the output of the source frame and the output of the target frame is within the error range. If it is not within the error range, it is necessary to locate which network layer of the neural network model causes different inference results under the two frameworks. At present, the open source neural network model conversion tools do not verify whether the output accuracy of the neural network model before and after the conversion is consistent. The user needs to manually run and save the results under the two frameworks, and then compare them. When the accuracy is inconsistent, the user needs to modify the neural network model based on experience, output the results of some layers in the modified neural network model, and locate the network layer that makes the inference results under the two frameworks inconsistent. This method will take a lot of time, and for some relatively large neural network models, it is often very difficult to locate the problem.

Please refer to FIG. 5, which is a schematic flowchart of a model accuracy positioning method provided by an embodiment of the present invention. The method includes the following steps:

S301: The model accuracy positioning device acquires the first neural network model under the source frame, the model input data, and the information of the target frame.

The model accuracy positioning device in this embodiment is only the difference in name from the model conversion device and neural network model conversion tool in the foregoing embodiment, and the same function can be realized. In the case where the accuracy of the second neural network model is known to be misaligned with the first neural network model, the model accuracy positioning device can perform model accuracy positioning based on a user instruction, which is used to instruct the positioning of the second neural network The target network layer in the road model with misaligned accuracy, the user command includes the first neural network model under the source frame, model input data, and information about the target frame. For example, when the user obtains the result that the accuracy of the second neural network model and the first neural network model are not aligned, the user instruction is sent to the model accuracy positioning device. The model accuracy positioning device can also perform model accuracy positioning based on internal operating instructions, such as model accuracy positioning instructions. For example, when the model accuracy positioning device obtains an instruction that the accuracy of the second neural network model and the first neural network model are not aligned, it triggers the model accuracy positioning instruction. The internal operation instructions are used to instruct to locate the target network layer of the second neural network model whose accuracy is not aligned. The internal operation instructions include the first neural network model under the source frame, model input data, and information about the target frame .

For the specific implementation of this step, refer to step S101 in the embodiment shown in FIG. 1 or step S201 in the embodiment shown in FIG. 3.

S302. The model accuracy positioning device performs model conversion processing on the first neural network model to obtain a second neural network model under the target framework.

For the specific implementation of this step, refer to step S102 in the embodiment shown in FIG. 1 or step S202 in the embodiment shown in FIG. 3.

S303. The model accuracy positioning device processes the output data of each network layer of the model input data separately based on the first neural network model and the second neural network model, and locates the target network whose accuracy is not aligned in the second neural network model Floor.

In the process of performing the model conversion processing on the first neural network model, the model accuracy positioning device runs the inference process of the first neural network model on the input data of the model to obtain the first neural network model to process the input data of the model The output data of each network layer. Optionally, the output data of each network layer of the first neural network model can be stored or recorded locally.

In other embodiments, the model accuracy positioning device may also re-input the model input data to the first neural network model, and run the inference process of the first neural network model to the model input data to obtain the first neural network model-to-model The output data of each network layer where the input data is processed.

In other embodiments, the model accuracy positioning device may pre-store the output data of each network layer obtained by the first neural network model inferring the model input data.

After obtaining the second neural network model under the target framework, the above model input data can be input to the second neural network model, and the model input data can be used to run on each network layer of the second neural network model Inferring again, the inference result output by the previous network layer can be used as the input data of the next network layer to obtain the output data of each network layer in the second neural network model.

After obtaining the first inference result and the second inference result, because the first inference result and the second inference result are obtained by using the same model input data to run the inference in the first neural network model and the second neural network model, respectively Therefore, according to the first inference result and the second inference result, for example, by comparing the first inference result and the second inference result, the accuracy verification result of the second neural network model under the target framework can be obtained. For another example, multiple parameter values in the second inference result can be matched with the corresponding parameter values in the first inference result one by one to obtain a matching result, that is, the aforementioned accuracy verification result can be obtained. In some embodiments, the accuracy verification result may indicate a comparison result of the model accuracy of the second neural network model and the first neural network model, such as accuracy alignment or accuracy misalignment, or the accuracy verification result may indicate the second neural network model. Whether the model accuracy of the network model meets expectations or a set standard, for example, the accuracy meets the standard or the accuracy does not meet the standard, etc., the embodiment of the present invention does not limit this.

After obtaining the output results of each network layer of the first neural network model and the output results of each network layer of the second neural network model, due to the The output results and the output results of each network layer of the second neural network model are obtained by using the same model input data to run inferences in the first neural network model and the second neural network model. Therefore, according to the first neural network model The output result of each network layer of a neural network model and the output result of each network layer of the second neural network model can locate the target network layer whose accuracy is not aligned in the second neural network model.

In one implementation, the target network layer is the network layer in which accuracy misalignment appears first among the multiple network layers included in the second neural network model. Specifically, if the difference between the output data of the network layer included in the second neural network model and the corresponding network layer of the first neural network model exceeds the preset error range, the network layer is deemed to be misaligned in accuracy Network layer.

In another implementation, the target network layer is a plurality of network layers with misaligned accuracy in the second neural network model. For example, it includes not only the first network layer where accuracy misalignment occurs, but also the subsequent network layers of the network layer. The number of subsequent network layers can be preset.

Specifically, sequentially compare the inference results of each network layer in the first neural network model under the source framework with the inference results of the corresponding network layer in the second neural network model under the target framework, and locate the second In the neural network model, the network layer whose accuracy is not aligned with the first neural network model first is the target network layer.

S304. The model accuracy positioning device outputs the information of the target network layer.

After determining the target network layer, the model accuracy positioning device can output the information of the target network layer. The information of the target network layer includes the identification, parameters, output data of the target network layer, and the error between the output data and the output data of the network layer corresponding to the target network layer in the first neural network model Wait. The administrator or user can analyze the reason for the misalignment of accuracy based on the information of the target network layer.

According to a model accuracy positioning method provided by an embodiment of the present invention, the output of each network layer obtained from the input data of the same model according to the first neural network model under the source frame and the second neural network model under the target frame The data can be used to locate the target network layer whose accuracy is not aligned in the second neural network model, and realize the accurate positioning of the target network layer when the accuracy is not aligned.

Please refer to FIG. 6, which is a schematic flowchart of another model accuracy positioning method provided by an embodiment of the present invention. The method includes the following steps:

S401: The model accuracy positioning device acquires the first neural network model under the source frame, the model input data, and the information of the target frame.

For the specific implementation of this step, refer to step S101 in the embodiment shown in FIG. 1, or step S201 in the embodiment shown in FIG. 3, or step S301 in the embodiment shown in FIG. 5.

S402. The model accuracy positioning device creates a network topology map by processing the input data of the model by the first neural network model, where the network topology map includes the information of the network layer in the first neural network model and the network Layer connection relationship.

The model accuracy positioning device creates a network topology map through the processing of the input data of the model by the first neural network model, wherein the network topology map includes the information of the network layer in the first neural network model and the network layer information Connection relationship. In the process of using the neural network model conversion tool to convert each network layer of the first neural network model under the source framework into the format of the target framework, the source framework will use the model to input data on the first neural network model Run the reasoning again to obtain the information of the network layer in the first neural network model and the connection relationship of the network layer. The information of the network layer in the above-mentioned first neural network model and the connection relationship between the network layers constitute the network topology diagram of the first neural network model.

S403. The model accuracy positioning device obtains the second neural network model based on the network topology map and the information of the target frame, and records the output data of each network layer in the process of processing the input data of the model by the first neural network model, and Based on the model conversion processing, a mapping relationship between the network layers in the first neural network model and the second neural network model is established.

After the model accuracy positioning device obtains the network topology of the first neural network, the target framework has its own network structure definition and toolkit. The model accuracy positioning device can be based on the network topology of the first neural network and use the target The network structure definition and toolkit of the framework adaptively configure the network topology of the first neural network to obtain the second neural network model.

Since the neural network model conversion tool is used to convert each network layer of the first neural network model under the source framework into the format of the target framework, the source framework will use the model to input data in the first neural network model. Run the reasoning on it once to obtain the information of the network layer in the first neural network model and the connection relationship of the network layer. Then, the model accuracy positioning device can be based on the network topology of the first neural network, using the network structure definition of the target framework and the toolkit to adaptively configure the network topology of the first neural network to obtain the second Neural network model. Therefore, further, in the process of model conversion processing, the mapping relationship between the network layers in the first neural network model and the second neural network model can also be established. Among them, the mapping relationship between the network layers of the first neural network model and the second neural network model can be one-to-many, many-to-one, or one-to-one. Accordingly, the target network layer can be one or more Network layer.

S404. The model accuracy positioning device processes the model input data through the second neural network model to obtain output data of each network layer in the second neural network model.

For the specific implementation of this step, refer to step S103 in the embodiment shown in FIG. 1 or step S203 in the embodiment shown in FIG. 3.

The difference is that in this embodiment, the output data of each network layer of the second neural network model is recorded, while in the embodiment shown in FIG. 1 or 3, the second neural network is recorded. The final inference result of the road model is the output data of the last network layer of the second neural network model.

S405. The model accuracy positioning device determines the target network layer in the second neural network model whose accuracy is not aligned based on the output data and the mapping relationship of each network layer in the first neural network model and the second neural network model.

The model accuracy positioning device obtains the output data of each network layer in the first neural network model and the second neural network model, and the difference between the network layers in the first neural network model and the second neural network model After the mapping relationship between the two, based on the output data and mapping relationship of each network layer in the first neural network model and the second neural network model, it can be determined that the target network with misaligned accuracy first appears in the second neural network model. Road layer.

Specifically, in the accuracy positioning process, in order to accurately locate the target network layer whose accuracy is not aligned first, a serial number is assigned to each network layer in the first neural network model. The sequence number can be in ascending powers, so that when positioning accuracy is a problem, each network layer in the first neural network model can be called in turn according to the sequence numbers of multiple network layers in the first neural network model.

For each network layer called, the inference result output by the previous network layer is input to the next network layer for inference, and the inference result of each network layer is obtained. Moreover, according to the serial number of each network layer, the inference result of each network layer is correspondingly stored locally. And also identify the inference results of each network layer. Among them, a network layer may have multiple inputs. Only save the inference results of each layer output, because in addition to the initial input data, the inputs of other layers are the output of the previous layer, so once the initial input of the first neural network model and the output of each layer are recorded, it is equivalent to recording The input and output of all layers are included. As shown in Figure 4, it is a schematic diagram of the positioning process of the target network layer when the accuracy is not aligned after the conversion of the example neural network model. The neural network model includes multiple network layers: convolutional layer, activation function, and normalization. Layer, Eltwise layer and pooling layer, etc. After the initial input data to the first network layer of the neural network model, the output of each network layer is used as the input of the next network layer. The inference results of the neural network model include initial input data and input/output data 1-10.

Each network layer in the first neural network model is individually converted to the target framework to obtain the second neural network model. Input the same model input data to the second neural network model, and then use the input of each layer to run the inference results of each layer of the model in the second neural network model, and save it locally. The serial number is the index. Of course, the reasoning under the target framework also needs to be carried out in a new process.

After obtaining the inference results of each network layer of the neural network model in the source frame and the target frame, you can compare each layer in the first neural network model in the source frame according to the sequence number of each network layer. The inference result of a network layer and the inference result of the corresponding network layer in the second neural network model under the target framework are located, and the network in the second neural network model whose accuracy is not aligned with the first neural network model first Road layer, that is, the target network layer.

S406: The model accuracy positioning device outputs the information of the target network layer.

For the specific implementation of this step, refer to step S304 of the embodiment shown in FIG. 5.

According to a model accuracy positioning method provided by an embodiment of the present invention, the output of each network layer obtained from the input data of the same model according to the first neural network model under the source frame and the second neural network model under the target frame The data can locate the target network layer whose accuracy is not aligned in the second neural network model, and realize the accurate positioning of the network layer when the accuracy is not aligned.

Based on the same concept of the above neural network model conversion method, as shown in FIG. 7, an embodiment of the present invention also provides a model conversion device, and the device 1000 includes:

The obtaining unit 11 is used to obtain the first neural network model under the source frame, the model input data, and the information of the target frame;

The conversion unit 12 is configured to perform model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework;

The verification unit 13 is configured to obtain a first inference result and a second inference result obtained by processing the model input data based on the first neural network model and the second neural network model, respectively, to obtain the second The accuracy verification result of the neural network model.

In one implementation, the conversion unit 12 is configured to perform a model conversion process on the first neural network model under the source framework through a first process to obtain the second neural network model under the target framework;

The device also includes:

The creating unit 14 is configured to create a second process through the first process;

The inference unit 15 is configured to invoke the second process through the first process, so that the second process executes the processing of the second neural network model on the model input data to obtain a second inference result.

In another implementation, the verification unit 13 is configured to determine the accuracy verification of the second neural network model if the difference between the first inference result and the second inference result is within a preset error range The result is precision aligned; or

The verification unit 13 is further configured to determine that the accuracy verification result of the second neural network model is not accurate if the difference between the first inference result and the second inference result exceeds the preset error range. Aligned.

In yet another implementation, the device further includes:

The positioning unit 16 is configured to locate the target network layer in the second neural network model whose accuracy is not aligned in response to determining that the accuracy verification result of the second neural network model is accuracy misalignment;

The output unit 17 is used to output the information of the target network layer.

In another implementation, the positioning unit 16 includes:

The establishing unit 161 is configured to establish a mapping relationship between the network layers in the first neural network model and the second neural network model;

The determining unit 162 is configured to determine the output data of each network layer and the mapping relationship in the process of separately processing the input data of the model based on the first neural network model and the second neural network model The target network layer whose accuracy is not aligned in the second neural network model.

In yet another implementation, wherein the model conversion device performs a model conversion process on the first neural network model again, wherein, in the process of performing the model conversion process again, the first neural network is obtained Output data of each network layer in the model, and/or establish a mapping relationship between the network layers in the first neural network model and the second neural network model.

In another implementation, the conversion unit 12 is used to process the input data of the model through the first neural network model to establish a network topology diagram, wherein the network topology diagram includes the first neural network model. The information of the network layer in the neural network model and the connection relationship of the network layer; and the second neural network model is obtained based on the network topology map and the information of the target frame.

For the specific implementation of the above-mentioned units, reference may be made to the relevant description in the neural network model conversion method shown in FIG. 1 or FIG. 3, which will not be repeated here.

According to a model conversion device provided by an embodiment of the present invention, the second neural network model in the source framework and the second neural network model in the target framework respectively obtain the inference results of the same model input data, and the second neural network model can be automatically obtained. The accuracy verification result of the neural network model improves the accuracy verification efficiency of the neural network model conversion.

Based on the same concept of the above model accuracy positioning method, as shown in FIG. 8, an embodiment of the present invention also provides a model accuracy positioning device. The device 2000 includes:

The obtaining unit 21 is used to obtain the first neural network model under the source frame, the model input data, and the information of the target frame;

The conversion unit 22 is configured to perform model conversion processing on the first neural network model to obtain a second neural network model under the target framework;

The positioning unit 23 is configured to locate the second neural network based on the output data of each network layer in the process of separately processing the input data of the model based on the first neural network model and the second neural network model Target network layer whose accuracy is not aligned in the road model;

The output unit 24 is used to output the information of the target network layer.

In one implementation, the information of the target network layer includes at least one of the following: the identification, parameters, output data of the target network layer, and the output data and the first neural network model. State the error between the output data of the network layer corresponding to the target network layer.

In yet another implementation, the device further includes:

The first establishing unit 25 is configured to establish a mapping relationship between the network layers in the first neural network model and the second neural network model based on the model conversion processing;

The positioning unit 23 is configured to determine the second neural network model based on the output data of each network layer in the first neural network model and the second neural network model and the mapping relationship The target network layer with medium precision misalignment.

In another implementation, the positioning unit 23 includes:

The calling unit 231 is configured to sequentially compare the output data of the corresponding network layer in the first neural network model and the second neural network model based on the mapping relationship;

The determining unit 232 is configured to respond to the difference between the output data of the corresponding network layer in the first neural network model and the second neural network model exceeding a preset error range, and determine the corresponding network layer The network layer in the second neural network model is determined as the target network layer.

In another implementation, the conversion unit 22 includes:

The second creating unit 221 is used to process the input data of the model by the first neural network model to create a network topology diagram, wherein the network topology diagram includes the first neural network model The information of the network layer and the connection relationship of the network layer;

The second obtaining unit 222 is configured to obtain the second neural network model based on the network topology map and the information of the target framework.

For the specific implementation of the foregoing units, reference may be made to the specific description in the embodiment shown in FIG. 5 or FIG. 6, which will not be repeated here.

According to a model accuracy positioning device provided by an embodiment of the present invention, the output of each network layer is obtained from the input data of the same model according to the first neural network model under the source frame and the second neural network model under the target frame The data can locate the target network layer whose accuracy is not aligned in the second neural network model, and realize the accurate positioning of the network layer when the accuracy is not aligned.

As shown in FIG. 9, a schematic diagram of the hardware structure of an electronic device is also provided, and the electronic device is used to execute the above-mentioned neural network model conversion method/model accuracy positioning method. Part or all of the above methods can be implemented by hardware, and can also be implemented by software or firmware.

Optionally, the electronic device may be a chip or an integrated circuit in specific implementation.

Optionally, when part or all of the neural network model conversion method/model accuracy positioning method of the foregoing embodiment is implemented by software or firmware, it can be implemented by an electronic device 3000 provided in FIG. 9. As shown in FIG. 9, the electronic device 3000 may include:

The memory 33 and the processor 34 (the processor 34 in the device may be one or more, and one processor is taken as an example in FIG. 9), and may also include an input device 31 and an output device 32. In this embodiment, the input device 31, the output device 32, the memory 33, and the processor 34 may be connected by a bus or other means, wherein the connection by a bus is taken as an example in FIG. 9.

Wherein, the processor 34 is used to execute the method steps executed in FIG. 1, FIG. 3, FIG. 5, and FIG. 6.

Optionally, the program of the above-mentioned neural network model conversion method/model accuracy positioning method may be stored in the memory 33. The memory 33 may be a physically independent unit, or may be integrated with the processor 34. The memory 33 can also be used to store data.

Optionally, when part or all of the neural network model conversion method/model accuracy positioning method of the foregoing embodiment is implemented by software, the electronic device may also only include a processor. The memory used to store the program is located outside the electronic device, and the processor is connected to the memory through a circuit or wire for reading and executing the program stored in the memory.

The processor may be a central processing unit (CPU), a network processor (NP), or a WLAN (Wireless Local Area Network) device.

The processor may further include a hardware chip. The aforementioned hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above-mentioned PLD can be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL) or any combination thereof .

The memory may include volatile memory (volatile memory), such as random-access memory (random-access memory, RAM); the memory may also include non-volatile memory (non-volatile memory), such as flash Memory (flash memory), hard disk drive (HDD) or solid-state drive (SSD); memory may also include a combination of the above types of memory.

Those skilled in the art should understand that one or more embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, one or more embodiments of the present invention may adopt the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware. Moreover, one or more embodiments of the present invention can be implemented on one or more computer usable storage media (including but not limited to magnetic disk memory, CD-ROM, optical memory, etc.) containing computer usable program codes. In the form of a computer program product.

The embodiment of the present invention also provides a computer-readable storage medium, the storage medium may store a computer program, and the program is executed by a processor to realize the steps of the neural network model conversion method described in any embodiment of the present invention. And/or, implement the steps of the model accuracy positioning method described in any embodiment of the present invention.

The various embodiments of the present invention are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the foregoing device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for related parts, please refer to the part of the description of the method embodiments.

The foregoing describes specific embodiments of the present invention. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than in the embodiments and still achieve desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown in order to achieve the desired results. In some embodiments, multiplexing and parallel processing are also possible or may be advantageous.

The embodiments of the subject and functional operations described in the present invention can be implemented in the following: digital electronic circuits, tangible computer software or firmware, computer hardware including the structure of the present invention and structural equivalents thereof, or their A combination of one or more of them. Embodiments of the subject matter described in the present invention can be implemented as one or more computer programs, that is, one or one of the computer program instructions encoded on a tangible non-transitory program carrier to be executed by a data processing device or to instruct the operation of the data processing device Multiple modules. Alternatively or additionally, the program instructions may be encoded on artificially generated propagated signals, such as machine-generated electrical, optical or electromagnetic signals, which are generated to encode information and transmit it to a suitable receiver device for data transmission. The processing device executes. The computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of one or more of them.

The processing and logic flow described in the present invention can be executed by one or more programmable computers executing one or more computer programs to perform corresponding functions by operating according to input data and generating output. The processing and logic flow can also be executed by a dedicated logic circuit, such as a field programmable gate array (FPGA) or a dedicated integrated circuit (ASIC), and the device can also be implemented as a dedicated logic circuit.

Computers suitable for executing computer programs include, for example, general-purpose and/or special-purpose microprocessors, or any other type of central processing unit. Generally, the central processing unit will receive commands and data from read-only memory and/or random access memory. The basic elements of a computer include a central processing unit for implementing or executing instructions and one or more memory devices for storing instructions and data. Generally, the computer will also include one or more large storage area devices for storing data, such as floppy disks, magneto-optical disks, or optical discs, or the computer will be operably coupled to this large storage area device to receive data or Send data to it, or both. However, the computer does not have to have such equipment. In addition, the computer can be embedded in another device, such as a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver, or a universal serial bus (USB) ) Portable storage devices for flash drives, to name a few.

Computer-readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media, and memory devices, such as semiconductor memory devices (such as EPROM, EEPROM, and flash memory devices), magnetic Disks (such as internal hard disks or removable disks), magneto-optical disks, and CD ROM and DVD-ROM disks. The processor and memory can be supplemented by or incorporated into a dedicated logic circuit.

Although the present invention contains many specific implementation details, these should not be construed as limiting the scope of any invention or the claimed scope, but are mainly used to describe the features of specific embodiments of the specific invention. Certain features described in multiple embodiments of the present invention can also be implemented in combination in a single embodiment. On the other hand, various features described in a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub-combination. In addition, although features may function in certain combinations as described above and even initially claimed as such, one or more features from the claimed combination may in some cases be removed from the combination, and the claimed The combination of protection can be directed to a sub-combination or a variant of the sub-combination.

Similarly, although operations are depicted in a specific order in the drawings, this should not be construed as requiring these operations to be performed in the specific order shown or sequentially, or requiring all illustrated operations to be performed to achieve desired results . In some cases, multiplexing and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can usually be integrated together in a single software product. Or packaged into multiple software products.

Thus, specific embodiments of the subject matter have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve desired results. In addition, the processes depicted in the drawings are not necessarily in the specific order or sequential order shown in order to achieve the desired result. In some implementations, multiplexing and parallel processing may be advantageous.

The foregoing descriptions are only preferred embodiments of one or more embodiments of the present invention, and are not intended to limit one or more embodiments of the present invention. All within the spirit and principle of one or more embodiments of the present invention, Any modifications, equivalent replacements, improvements, etc. made should be included in the protection scope of one or more embodiments of the present invention.

1000: Model conversion device 11, 21: get unit 12, 22: conversion unit 13: Verification unit 14: Create a unit 15: Reasoning Unit 16, 23: positioning unit 161: Building Unit 162, 232: Determining the unit 17, 24: output unit 221: The second establishment unit 222: The second acquisition unit 231: call unit 25: The first establishment unit 2000: Model accuracy positioning device 3000: electronic equipment 31: Input device 32: output device 33: Memory 34: processor S101～S103, S201～S209, S301～S304, S401～S406: steps

FIG. 1 is a schematic flowchart of a neural network model conversion method provided by an embodiment of the present invention; 2 is a schematic diagram of a conversion process of a neural network model provided by an embodiment of the present invention; 3 is a schematic diagram of another process of the neural network model conversion method provided by the embodiment of the present invention; FIG. 4 is a schematic diagram of a network topology obtained during a target network layer positioning process provided by an embodiment of the present invention; 5 is a schematic flowchart of a model accuracy positioning method provided by an embodiment of the present invention; FIG. 6 is another schematic flowchart of a model accuracy positioning method provided by an embodiment of the present invention; FIG. 7 is a schematic structural diagram of a model conversion device provided by an embodiment of the present invention; FIG. 8 is a schematic structural diagram of a model accuracy positioning device provided by an embodiment of the present invention; FIG. 9 is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention.

S101~S103: steps

Claims (15)

A neural network model conversion method, characterized in that the method includes: The model conversion device obtains the first neural network model under the source frame, model input data, and information about the target frame; The model conversion device performs model conversion processing on the first neural network model in the source frame to obtain the second neural network model in the target frame; The model conversion device obtains the second neural network model based on the first inference result and the second inference result obtained by processing the model input data respectively based on the first neural network model and the second neural network model The accuracy verification result of the network model. The method according to claim 1, wherein performing the model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework includes: Performing model conversion processing on the first neural network model under the source framework through the first process to obtain the second neural network model under the target framework; The method also includes: Creating a second process through the first process; Invoke the second process through the first process to process the model input data through the second neural network model in the second process to obtain a second inference result. The method according to claim 1 or 2, wherein the model conversion device is based on the first neural network model and the second neural network model to process the model input data to obtain the first An inference result and a second inference result to obtain the accuracy verification result of the second neural network model, including: If the difference between the first inference result and the second inference result is within a preset error range, determining that the accuracy verification result of the second neural network model is accuracy alignment; or If the difference between the first inference result and the second inference result exceeds the preset error range, it is determined that the accuracy verification result of the second neural network model is accuracy misalignment. The method according to any one of claims 1 to 3, wherein the method further includes: In response to determining that the accuracy verification result of the second neural network model is accuracy misalignment, the model conversion device locates a target network layer whose accuracy is not aligned in the second neural network model; The model conversion device outputs the information of the target network layer. The method according to claim 4, wherein the locating a target network layer whose accuracy is not aligned in the second neural network model includes: The model conversion device establishes a mapping relationship between the network layers in the first neural network model and the second neural network model; The model conversion device determines the output data of each network layer and the mapping relationship in the process of separately processing the input data of the model based on the first neural network model and the second neural network model The target network layer whose accuracy is not aligned in the second neural network model. The method according to claim 5, wherein the model conversion device performs a model conversion process on the first neural network model again, wherein, in the process of performing the model conversion process again, obtain Output data of each network layer in the first neural network model, and/or establish a mapping relationship between the network layers in the first neural network model and the second neural network model . The method according to any one of claims 1 to 6, wherein the model conversion device performs model conversion processing on the first neural network model in the source framework to obtain the first neural network model in the target framework Two neural network models, including: The model conversion device creates a network topology map by processing the input data of the model by the first neural network model, wherein the network topology map includes the network in the first neural network model Layer information and the connection relationship of the network layer; The model conversion device obtains the second neural network model based on the network topology map and the information of the target framework. A model accuracy positioning method, which is characterized in that it includes: The model accuracy positioning device acquires the first neural network model under the source frame, model input data, and information about the target frame; The model accuracy positioning device performs model conversion processing on the first neural network model to obtain a second neural network model under the target framework; The model accuracy positioning device locates the second neural network based on the output data of each network layer in the process of separately processing the input data of the model by the first neural network model and the second neural network model Target network layer whose accuracy is not aligned in the road model; The model accuracy positioning device outputs the information of the target network layer. The method according to claim 8, wherein the method further includes: The model accuracy positioning device establishes a mapping relationship between the network layers in the first neural network model and the second neural network model based on the model conversion processing; The model accuracy positioning device locates the second neural network based on the output data of each network layer in the process of separately processing the input data of the model by the first neural network model and the second neural network model The target network layer whose accuracy is not aligned in the road model includes: The model accuracy positioning device determines the accuracy in the second neural network model based on the output data of each network layer in the first neural network model and the second neural network model and the mapping relationship Unaligned target network layer. The method according to claim 9, wherein the model accuracy positioning device is based on the output data of each network layer in the first neural network model and the second neural network model and the mapping Relationship, determining the target network layer whose accuracy is not aligned in the second neural network model, including: The model accuracy positioning device sequentially compares output data of corresponding network layers in the first neural network model and the second neural network model based on the mapping relationship; In response to the difference between the output data of the corresponding network layer in the first neural network model and the second neural network model exceeding a preset error range, classify the corresponding network layer as belonging to the second The network layer in the neural network model is determined to be the target network layer. The method according to any one of claim 8 to 10, wherein the model accuracy positioning device performs model conversion processing on the first neural network model under the source framework to obtain the The second neural network model includes: The model accuracy positioning device creates a network topology map by processing the input data of the model by the first neural network model, wherein the network topology map includes the network in the first neural network model The information of the road layer and the connection relationship of the network layer; The model accuracy positioning device obtains the second neural network model based on the network topology map and the information of the target frame. A model conversion device, characterized in that the device includes: The obtaining unit is used to obtain the first neural network model under the source frame, the model input data, and the information of the target frame; A conversion unit, configured to perform model conversion processing on the first neural network model in the source framework to obtain the second neural network model in the target framework; The verification unit is configured to obtain the first inference result and the second inference result obtained by processing the input data of the model based on the first neural network model and the second neural network model, respectively, to obtain the second neural network The accuracy verification result of the network model. A model accuracy positioning device, which is characterized in that it comprises: The obtaining unit is used to obtain the first neural network model under the source frame, the model input data, and the information of the target frame; A conversion unit for performing model conversion processing on the first neural network model to obtain a second neural network model under the target framework; The positioning unit is configured to locate the second neural network based on the output data of each network layer in the process of separately processing the input data of the model based on the first neural network model and the second neural network model Target network layer whose accuracy is not aligned in the model; The output unit is used to output the information of the target network layer. An electronic device, characterized in that the device includes a memory and a processor, wherein the memory stores program instructions, and the processor is used to call the program instructions stored in the memory to execute as claimed The method described in any one of 1 to 11. A computer storage medium, characterized in that instructions are stored in the computer-readable storage medium, which when run on a computer, cause the computer to execute the method according to any one of claims 1 to 11.

Images