File capture and configuration method and system for algorithm model conversion

Abstract

The present invention provides a file capture and configuration method and system for algorithm model conversion. Based on the conversion progress information of the algorithm model on the platform side, the file attribute information required by the algorithm model in the conversion process is predicted, so that a file capture broadcast message is sent to all server sides to determine a suitable server side; file data is captured from a suitable server side and a corresponding file data packet is formed, and the file data packet matching the conversion of the algorithm model is screened and added to a queue of configuration files to be configured, so as to provide reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, the corresponding file is extracted from the queue of configuration files to be configured, and code data matching the conversion of the algorithm model is generated, so as to perform code-level conversion configuration on the algorithm model, and based on the conversion configuration result, the processing status of the algorithm model on the platform side is adjusted, and a unified and standardized code configuration is performed on the conversion of the algorithm model, so as to improve the conversion success rate and reliability of the algorithm model.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a file capture and configuration method and system for algorithm model conversion. Background Art

[0002] In order to adapt to the computing requirements of different occasions, the algorithm model needs to be converted before it can process actual computing tasks. The conversion of the algorithm model involves editing the internal code of the algorithm model, which requires users to have corresponding code writing experience to achieve this, which increases the difficulty and cost of the algorithm model conversion. In addition, there are differences in the specific codes written by different programmers during the algorithm model conversion process, and it is impossible to perform a unified and standardized code configuration for the conversion of the algorithm model, which reduces the conversion success and reliability of the algorithm model, and cannot guarantee the efficiency and accuracy of the algorithm model conversion. Summary of the invention

[0003] The purpose of the present invention is to provide a file capture and configuration method and system for algorithm model conversion, which predicts the file attribute information required by the algorithm model in the conversion process based on the conversion progress information of the algorithm model on the platform side, and sends a file capture broadcast message to all server sides to determine the appropriate server side; captures file data from the appropriate server side and forms a corresponding file data packet, screens the file data packets that match the conversion of the algorithm model and adds them to the queue of configuration files to be configured, so as to provide reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, extracts the corresponding files from the queue of configuration files to be configured, generates code data that matches the conversion of the algorithm model, so as to perform code-level conversion configuration of the algorithm model, and also adjusts the processing status of the algorithm model on the platform side based on the conversion configuration result, performs unified and standardized code configuration for the conversion of the algorithm model, improves the conversion success rate and reliability of the algorithm model, and ensures the efficiency and accuracy of the algorithm model conversion.

[0004] The present invention is achieved through the following technical solutions:

[0005] File capture and configuration methods for algorithm model conversion, including:

[0006] Acquire the conversion progress information of the algorithm model on the platform end, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process; based on the file attribute information, send a file capture broadcast message to all the servers connected to the platform end, so as to determine all the servers that can provide the corresponding files;

[0007] Based on the respective file status information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; the file data packet received by the platform is identified and processed, the file data packet matching the conversion operation of the algorithm model is screened, and the screened file data packet is added to the queue of the to-be-configured files;

[0008] Based on the conversion log of the algorithm model, the corresponding files are extracted from the queue of configuration files, and the extracted files are parsed and processed to generate code data that matches the conversion of the algorithm model, so as to convert and configure the algorithm model; based on the conversion configuration result of the algorithm model, the processing status of the algorithm model on the platform end is adjusted.

[0009] Optionally, the conversion progress information of the algorithm model on the platform side is obtained, and based on the conversion progress information, the file attribute information required by the algorithm model during the conversion process is predicted; based on the file attribute information, a file crawling broadcast message is sent to all servers connected to the platform side, so as to determine all servers that can provide corresponding files, including:

[0010] Monitor the algorithm model conversion program on the platform end to obtain conversion progress information of the algorithm model conversion task performed by the algorithm model conversion program; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process;

[0011] Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directories returned by all servers regarding the file capture broadcast message are identified to determine whether the server contains the files required for the algorithm model conversion; if so, the corresponding server is determined as a server that can provide the corresponding files; if not, the corresponding server is determined as a server that cannot provide the corresponding files.

[0012] Optionally, based on the respective file status information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; the file data packet received by the platform is identified and processed, the file data packet matching the conversion operation of the algorithm model is screened, and the screened file data packet is added to the queue of the to-be-configured files, including:

[0013] Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end;

[0014] Perform file data integrity identification processing on the file data packet received by the platform end to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

[0015] Optionally, compressing the captured file data into a file data packet and returning it to the platform end includes:

[0016] Step S1, using the following formula (1), the captured file data is first compressed.

[0017] g1 16 =[Y 16 < <len(G 16 )]-(1< <Y 16 ) (1)

[0018] In the above formula (1), g1 16 Indicates that the captured file data is compressed in hexadecimal format; g 16 represents the hexadecimal form of the fixed information data of the platform end; len() represents the total number of bits of the hexadecimal data in the brackets; G 16 Indicates the hexadecimal form of the captured file data; << indicates left shift;

[0019] Step S2, using the following formula (2), the captured file data is subjected to a second compression.

[0020] g2 16 =[G 16 < <len(Y 16 )]-(Y 16 <<1) (2)

[0021] In the above formula (2), g2 16 Indicates that the captured file data is compressed in hexadecimal format;

[0022] The first compressed hexadecimal data is sent first, and then the second compressed hexadecimal data is sent;

[0023] Step S3, when the platform receives the first compressed and second compressed hexadecimal data, the platform uses the following formula (3) to decompress and verify the first compressed and second compressed hexadecimal data.

[0024]

[0025] In the above formula (3), G1 16 Indicates the hexadecimal form after decompression of the first compressed hexadecimal data received; G2 16 Indicates the hexadecimal form after decompression of the second compressed hexadecimal data received; g1′ 16 Indicates the first compressed hexadecimal data received; g2′ 16 Indicates the second compressed hexadecimal data received; >> indicates right shift; Indicates circular left shift; F() indicates a judgment function. If the formula in the brackets is true, the function value of the judgment function is 1. If the formula in the brackets is not true, the function value of the judgment function is 0; E indicates a check value;

[0026] If E=1, it means the verification is successful;

[0027] If E=0, it means that the verification fails, and the captured file data needs to be compressed again to form a file data packet and then returned to the platform end.

[0028] Optionally, based on the conversion log of the algorithm model, extracting corresponding files from the queue of configuration files, parsing the extracted files, generating code data matching the conversion of the algorithm model, and thereby converting and configuring the algorithm model; based on the conversion configuration result of the algorithm model, adjusting the processing status of the algorithm model by the platform end, including:

[0029] Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; parse the extracted file to generate code data that matches the conversion of the algorithm model, and record the code data in the algorithm model, so as to convert and configure the algorithm model;

[0030] Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

[0031] File capture and configuration system for algorithm model conversion, including:

[0032] A file attribute determination module, used to obtain conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process;

[0033] A file search module, used to send a file capture broadcast message to all servers connected to the platform end based on the file attribute information, so as to determine all servers that can provide the corresponding file;

[0034] A file capture processing module, for capturing corresponding file data from each server end based on the file status information of each server end that can provide the corresponding file, and returning the corresponding file data packet to the platform end;

[0035] A file data screening and processing module, used to identify and process file data packets received by the platform, screen file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to a queue for configuration files;

[0036] An algorithm model conversion configuration module is used to extract corresponding files from the queue of configuration files based on the conversion log of the algorithm model, parse the extracted files, generate code data matching the conversion of the algorithm model, and perform conversion configuration on the algorithm model;

[0037] The algorithm model processing adjustment module is used to adjust the processing status of the algorithm model by the platform end based on the conversion configuration result of the algorithm model.

[0038] Optionally, the file attribute determination module is used to obtain conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process, including:

[0039] Monitor the algorithm model conversion program on the platform end to obtain the conversion progress information of the algorithm model conversion program executing the algorithm model conversion task; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process;

[0040] The file search module is used to send a file capture broadcast message to all servers connected to the platform end based on the file attribute information, so as to determine all servers that can provide the corresponding file, including:

[0041] Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directories returned by all servers regarding the file capture broadcast message are identified to determine whether the server contains the files required for the algorithm model conversion; if so, the corresponding server is determined as a server that can provide the corresponding files; if not, the corresponding server is determined as a server that cannot provide the corresponding files.

[0042] Optionally, the file capture processing module is used to capture corresponding file data from each server end based on the file status information of each server end that can provide the corresponding file, and return the corresponding file data packet to the platform end, including:

[0043] Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end;

[0044] The file data screening and processing module is used to identify and process the file data packets received by the platform end, screen the file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to the queue for configuration files, including:

[0045] Perform file data integrity identification processing on the file data packet received by the platform end to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

[0046] Optionally, the algorithm model conversion configuration module is used to extract corresponding files from the queue of configuration files based on the conversion log of the algorithm model, and parse the extracted files to generate code data matching the conversion of the algorithm model, so as to perform conversion configuration on the algorithm model, including:

[0047] Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; parse the extracted file to generate code data that matches the conversion of the algorithm model, and record the code data in the algorithm model, so as to convert and configure the algorithm model;

[0048] The algorithm model processing adjustment module is used to adjust the processing state of the algorithm model by the platform end based on the conversion configuration result of the algorithm model, including:

[0049] Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] The file capture and configuration method and system for algorithm model conversion provided in the present application predict the file attribute information required by the algorithm model during the conversion process based on the conversion progress information of the algorithm model on the platform side, so as to send a file capture broadcast message to all server sides to determine the appropriate server side; capture file data from the appropriate server side and form a corresponding file data packet, screen the file data packet that matches the conversion of the algorithm model and add it to the queue of configuration files to be configured, so as to provide reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, extract the corresponding file from the queue of configuration files to be configured, generate code data that matches the conversion of the algorithm model, so as to perform code-level conversion configuration of the algorithm model, and also based on the conversion configuration result, adjust the processing status of the algorithm model on the platform side, perform unified and standardized code configuration for the conversion of the algorithm model, improve the conversion success rate and reliability of the algorithm model, and ensure the efficiency and accuracy of the algorithm model conversion. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0053] Figure 1 A schematic flow chart of a file capture and configuration method for algorithm model conversion provided by the present invention.

[0054] Figure 2 A structural diagram of a file capture and configuration system for algorithm model conversion provided by the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some structures related to the present application are shown in the accompanying drawings, rather than all structures. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0056] The terms "including" and "having" and any variations thereof in this application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0057] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0058] See also Figure 1 As shown, the file capture and configuration method for algorithm model conversion provided by an embodiment of the present application includes:

[0059] Acquire the conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process; based on the file attribute information, send a file crawling broadcast message to all servers connected to the platform side, so as to determine all servers that can provide the corresponding files;

[0060] Based on the file status information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; the file data packet received by the platform is identified and processed, and the file data packet matching the conversion operation of the algorithm model is screened, and the screened file data packet is added to the queue of the configuration files to be configured;

[0061] Based on the conversion log of the algorithm model, the corresponding files are extracted from the queue of configuration files, and the extracted files are parsed and processed to generate code data that matches the conversion of the algorithm model, so as to convert and configure the algorithm model; based on the conversion configuration result of the algorithm model, the processing status of the algorithm model on the platform is adjusted.

[0062] The beneficial effects of the above embodiments are as follows: the file capture and configuration method for algorithm model conversion predicts the file attribute information required by the algorithm model during the conversion process based on the conversion progress information of the algorithm model on the platform side, thereby sending a file capture broadcast message to all server sides to determine the appropriate server side; captures file data from the appropriate server side and forms a corresponding file data packet, screens the file data packets that match the conversion of the algorithm model and adds them to the queue of configuration files to be configured, thereby providing reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, extracts the corresponding files from the queue of configuration files to be configured, generates code data that matches the conversion of the algorithm model, thereby performing code-level conversion configuration on the algorithm model, and based on the conversion configuration result, adjusts the processing status of the algorithm model on the platform side, performs unified and standardized code configuration on the conversion of the algorithm model, improves the conversion success rate and reliability of the algorithm model, and ensures the efficiency and accuracy of the algorithm model conversion.

[0063] In another embodiment, the conversion progress information of the algorithm model on the platform side is obtained, and based on the conversion progress information, the file attribute information required by the algorithm model in the conversion process is predicted; based on the file attribute information, a file crawling broadcast message is sent to all servers connected to the platform side, so as to determine all servers that can provide the corresponding files, including:

[0064] Monitor the algorithm model conversion program on the platform to obtain the conversion progress information of the algorithm model conversion task performed by the algorithm model conversion program; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process;

[0065] Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directory returned by each server regarding the file capture broadcast message is identified to determine whether the server contains the files required for the algorithm model conversion; if contained, the corresponding server is determined as a server that can provide the corresponding file; if not, the corresponding server is determined as a server that cannot provide the corresponding file.

[0066] The beneficial effects of the above embodiments are that the platform side, as a carrier for implementing the conversion of the algorithm model, provides the corresponding hardware and software conditions for the conversion of the algorithm model, monitors the algorithm model conversion program inside the platform side, and obtains the conversion progress information of the algorithm model conversion program executing the algorithm model conversion task. The conversion progress information is only the conversion stage information of the algorithm model conversion task; then the conversion progress information is compared with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process; the preset algorithm model conversion calendar refers to the algorithm model conversion benchmark process corresponding to the algorithm model in the conversion process. Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform side. After receiving the file capture broadcast message, each server will return its corresponding file directory to determine whether the server contains the files required for the algorithm model conversion, thereby accurately identifying the server that can provide the corresponding files and effectively narrowing the scope of subsequent file capture.

[0067] In another embodiment, based on the respective file status information of all servers that can provide the corresponding file, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; the file data packet received by the platform is identified and processed, the file data packet matching the conversion operation of the algorithm model is screened, and the screened file data packet is added to the queue of the configuration files to be configured, including:

[0068] Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end;

[0069] Perform file data integrity identification processing on the file data packet received by the platform to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

[0070] The beneficial effect of the above embodiment is that, based on the file storage location information of all the servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end, ensuring that the platform end has sufficient and reliable file data for algorithm model conversion configuration. In addition, the file data packet received by the platform end is processed for file data integrity identification to determine whether there is data missing in the file data packet, thereby ensuring that a comprehensive and complete file data packet is used for algorithm model conversion.

[0071] In another embodiment, the captured file data is compressed into a file data packet and then returned to the platform end, including:

[0072] Step S1, using the following formula (1), the captured file data is first compressed.

[0073] g1 16 =[Y 16 < <len(G 16 )]-(1< <Y 16 ) (1)

[0074] In the above formula (1), g1 16 Indicates that the captured file data is the hexadecimal data after the first compression; Y 16 Indicates the hexadecimal form of the fixed information data on the platform; len() indicates the total number of digits of the hexadecimal data in the brackets; G 16 Indicates the hexadecimal format of the captured file data; << indicates left shift;

[0075] Step S2, using the following formula (2), the captured file data is compressed for the second time.

[0076] g2 16 =[G 16 < <len(Y 16 )]-(Y 16 <<1) (2)

[0077] In the above formula (2), g2 16 Indicates the captured file data in hexadecimal format after the second compression;

[0078] The first compressed hexadecimal data is sent first, and then the second compressed hexadecimal data is sent;

[0079] Step S3, when the platform receives the first compressed and second compressed hexadecimal data, the platform uses the following formula (3) to decompress and verify the first compressed and second compressed hexadecimal data.

[0080]

[0081] In the above formula (3), G1 16 g2 represents the decompressed hexadecimal form of the first compressed hexadecimal data received; 16 Indicates the hexadecimal form after decompression of the second compressed hexadecimal data received; g1′ 16 Indicates the first compressed hexadecimal data received; g2′16 Indicates the second compressed hexadecimal data received; >> indicates right shift; Indicates circular left shift; F() indicates a judgment function. If the formula in the brackets is true, the function value of the judgment function is 1. If the formula in the brackets is not true, the function value of the judgment function is 0; E indicates a check value;

[0082] If E=1, it means the verification is successful;

[0083] If E=0, it means that the verification fails, and the captured file data needs to be compressed again to form a file data packet and then returned to the platform end.

[0084] The beneficial effects of the above embodiments are:

[0085] In another embodiment, based on the conversion log of the algorithm model, the corresponding file is extracted from the queue of configuration files, and the extracted file is parsed and processed to generate code data matching the conversion of the algorithm model, so as to convert and configure the algorithm model; based on the conversion configuration result of the algorithm model, the processing state of the algorithm model by the platform end is adjusted, including:

[0086] Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; parse the extracted file to generate code data that matches the conversion of the algorithm model, and record the code data in the algorithm model, so as to perform conversion configuration on the algorithm model;

[0087] Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

[0088] The beneficial effects of the above-mentioned embodiments include analyzing the conversion log of the algorithm model, determining the real-time conversion process of the algorithm model, extracting the corresponding file from the queue of files to be configured; parsing the extracted file, generating code data that matches the conversion of the algorithm model, and recording the code data in the algorithm model, thereby converting and configuring the algorithm model to achieve a unified and standardized conversion of the algorithm model. In addition, based on the conversion configuration result of the algorithm model, it is determined whether the algorithm model successfully executes the current conversion process. If so, the conversion process of the algorithm model on the platform side is kept unchanged; if not, the conversion process of the algorithm model on the platform side is re-executed, thereby improving the conversion success and reliability of the algorithm model.

[0089] See also Figure 2As shown, the file capture and configuration system for algorithm model conversion provided by an embodiment of the present application includes:

[0090] A file attribute determination module is used to obtain the conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process;

[0091] The file search module is used to send a file capture broadcast message to all servers connected to the platform based on the file attribute information, so as to determine all servers that can provide the corresponding file;

[0092] A file capture processing module, which is used to capture corresponding file data from each server based on the file status information of all servers that can provide corresponding files, and return the corresponding file data packet to the platform end;

[0093] A file data screening and processing module is used to identify and process file data packets received by the platform, screen file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to a queue for configuration files;

[0094] The algorithm model conversion configuration module is used to extract the corresponding file from the queue of configuration files based on the conversion log of the algorithm model, parse the extracted file, generate code data matching the conversion of the algorithm model, and perform conversion configuration on the algorithm model;

[0095] The algorithm model processing adjustment module is used to adjust the processing status of the algorithm model by the platform end based on the conversion configuration result of the algorithm model.

[0096] The beneficial effects of the above embodiments are as follows: the file capture and configuration system for algorithm model conversion predicts the file attribute information required by the algorithm model during the conversion process based on the conversion progress information of the algorithm model on the platform side, thereby sending file capture broadcast messages to all server sides to determine the appropriate server side; captures file data from the appropriate server side and forms corresponding file data packets, screens file data packets that match the conversion of the algorithm model and adds them to the queue for configuration files, providing reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, extracts the corresponding files from the queue for configuration files, generates code data that matches the conversion of the algorithm model, thereby performing code-level conversion configuration of the algorithm model, and also based on the conversion configuration result, adjusts the processing status of the algorithm model on the platform side, performs unified and standardized code configuration for the conversion of the algorithm model, improves the conversion success rate and reliability of the algorithm model, and ensures the efficiency and accuracy of the algorithm model conversion.

[0097] In another embodiment, the file attribute determination module is used to obtain conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process, including:

[0098] Monitor the algorithm model conversion program on the platform to obtain the conversion progress information of the algorithm model conversion task performed by the algorithm model conversion program; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process;

[0099] The file search module is used to send a file crawling broadcast message to all servers connected to the platform based on the file attribute information, so as to determine all servers that can provide the corresponding file, including:

[0100] Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directory returned by each server regarding the file capture broadcast message is identified to determine whether the server contains the files required for the algorithm model conversion; if contained, the corresponding server is determined as a server that can provide the corresponding file; if not, the corresponding server is determined as a server that cannot provide the corresponding file.

[0101] The beneficial effects of the above embodiments are that the platform side, as a carrier for implementing the conversion of the algorithm model, provides the corresponding hardware and software conditions for the conversion of the algorithm model, monitors the algorithm model conversion program inside the platform side, and obtains the conversion progress information of the algorithm model conversion program executing the algorithm model conversion task. The conversion progress information is only the conversion stage information of the algorithm model conversion task; then the conversion progress information is compared with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process; the preset algorithm model conversion calendar refers to the algorithm model conversion benchmark process corresponding to the algorithm model in the conversion process. Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform side. After receiving the file capture broadcast message, each server will return its corresponding file directory to determine whether the server contains the files required for the algorithm model conversion, thereby accurately identifying the server that can provide the corresponding files and effectively narrowing the scope of subsequent file capture.

[0102] In another embodiment, the file capture processing module is used to capture corresponding file data from each server end based on the file status information of each server end that can provide the corresponding file, and return the corresponding file data packet to the platform end, including:

[0103] Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end;

[0104] The file data screening and processing module is used to identify and process the file data packets received by the platform, screen the file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to the queue for configuration files, including:

[0105] Perform file data integrity identification processing on the file data packet received by the platform to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

[0106] The beneficial effect of the above embodiment is that, based on the file storage location information of all the servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end, ensuring that the platform end has sufficient and reliable file data for algorithm model conversion configuration. In addition, the file data packet received by the platform end is processed for file data integrity identification to determine whether there is data missing in the file data packet, thereby ensuring that a comprehensive and complete file data packet is used for algorithm model conversion.

[0107] In another embodiment, the algorithm model conversion configuration module is used to extract corresponding files from the queue of configuration files based on the conversion log of the algorithm model, and parse the extracted files to generate code data that matches the conversion of the algorithm model, so as to perform conversion configuration on the algorithm model, including:

[0108] Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; parse the extracted file to generate code data that matches the conversion of the algorithm model, and record the code data in the algorithm model, so as to perform conversion configuration on the algorithm model;

[0109] The algorithm model processing adjustment module is used to adjust the processing state of the algorithm model by the platform end based on the conversion configuration result of the algorithm model, including:

[0110] Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

[0111] The beneficial effects of the above-mentioned embodiments include analyzing the conversion log of the algorithm model, determining the real-time conversion process of the algorithm model, extracting the corresponding file from the queue of files to be configured; parsing the extracted file, generating code data that matches the conversion of the algorithm model, and recording the code data in the algorithm model, thereby converting and configuring the algorithm model to achieve a unified and standardized conversion of the algorithm model. In addition, based on the conversion configuration result of the algorithm model, it is determined whether the algorithm model successfully executes the current conversion process. If so, the conversion process of the algorithm model on the platform side is kept unchanged; if not, the conversion process of the algorithm model on the platform side is re-executed, thereby improving the conversion success and reliability of the algorithm model.

[0112] In general, the file capture and configuration method and system for algorithm model conversion predicts the file attribute information required by the algorithm model during the conversion process based on the conversion progress information of the algorithm model on the platform side, and sends file capture broadcast messages to all servers to determine the appropriate server; captures file data from the appropriate server and forms corresponding file data packets, screens file data packets that match the conversion of the algorithm model and adds them to the queue of configuration files to be configured, providing reliable file configuration support for the algorithm model conversion; based on the conversion log of the algorithm model, extracts the corresponding files from the queue of configuration files to be configured, generates code data that matches the conversion of the algorithm model, and configures the algorithm model at the code level; based on the conversion configuration result, adjusts the processing status of the algorithm model on the platform side, performs unified and standardized code configuration for the conversion of the algorithm model, improves the conversion success rate and reliability of the algorithm model, and ensures the efficiency and accuracy of the algorithm model conversion.

[0113] The above is only a specific implementation of the present invention, and any other improvements made based on the concept of the present invention are considered to be within the protection scope of the present invention.

Claims

1. A file capture and configuration method for algorithm model conversion, characterized in that: include: Acquire the conversion progress information of the algorithm model on the platform end, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process; based on the file attribute information, send a file capture broadcast message to all the servers connected to the platform end, so as to determine all the servers that can provide the corresponding files; Based on the file status information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; Identify and process the file data packets received by the platform end, filter the file data packets that match the conversion operation of the algorithm model, and add the filtered file data packets to the queue of configuration files to be configured; based on the conversion log of the algorithm model, extract the corresponding files from the queue of configuration files to be configured, parse the extracted files, generate code data that matches the conversion of the algorithm model, and perform conversion configuration on the algorithm model; based on the conversion configuration result of the algorithm model, adjust the processing status of the algorithm model by the platform end; Among them, obtaining the conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predicting the file attribute information required by the algorithm model in the conversion process; based on the file attribute information, sending a file crawling broadcast message to all the servers connected to the platform side, so as to determine all the servers that can provide the corresponding files, including: Monitor the algorithm model conversion program on the platform end to obtain conversion progress information of the algorithm model conversion task performed by the algorithm model conversion program; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process; Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directories returned by all servers regarding the file capture broadcast message are identified to determine whether the server contains the files required for the algorithm model conversion; if so, the corresponding server is determined as a server that can provide the corresponding files; if not, the corresponding server is determined as a server that cannot provide the corresponding files.

2. The file capture and configuration method for algorithm model conversion according to claim 1, characterized in that: Based on the file status information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the corresponding file data packet is returned to the platform; Identifying and processing the file data packets received by the platform, screening the file data packets matching the conversion operation of the algorithm model, and adding the screened file data packets to the queue of configuration files to be configured, including: Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end; Perform file data integrity identification processing on the file data packet received by the platform end to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

3. The file capture and configuration method for algorithm model conversion according to claim 2, characterized in that: Compressing the captured file data into a file data packet and returning it to the platform end includes: Step S1, using the following formula (1), the captured file data is first compressed. g1 16 =[And 16 < <len(G 16 )]-(1< <Y 16 ) (1) In the above formula (1), g1 16 Indicates that the captured file data is compressed in hexadecimal format; Y 16 represents the hexadecimal form of the fixed information data of the platform end; len() represents the total number of bits of the hexadecimal data in the brackets; G 16 Indicates the hexadecimal form of the captured file data; << indicates left shift; Step S2, using the following formula (2), the captured file data is subjected to a second compression. g2 16 =[G 16 <<len(Y 16 )]-(Y 16 <<1) (2) In the above formula (2), g2 16 Indicates that the captured file data is compressed in hexadecimal format; The first compressed hexadecimal data is sent first, and then the second compressed hexadecimal data is sent; Step S3, when the platform receives the first compressed and second compressed hexadecimal data, the platform uses the following formula (3) to decompress and verify the first compressed and second compressed hexadecimal data. In the above formula (3), G1 16 G2 represents the hexadecimal form after decompression of the first compressed hexadecimal data received; 16 Indicates the hexadecimal form after decompression of the second compressed hexadecimal data received; g1′ 16 Indicates the first compressed hexadecimal data received; g2′ 16 Indicates the second compressed hexadecimal data received; >> indicates right shift; Indicates circular left shift; F() indicates a judgment function. If the formula in the brackets is true, the function value of the judgment function is 1. If the formula in the brackets is not true, the function value of the judgment function is 0; E indicates a check value; If E=1, it means the verification is successful; If E=0, it means that the verification fails, and the captured file data needs to be compressed again to form a file data packet and then returned to the platform end.

4. The file capture and configuration method for algorithm model conversion according to claim 1, characterized in that: Based on the conversion log of the algorithm model, extract the corresponding file from the queue of configuration files to be configured, parse the extracted file, generate code data matching the conversion of the algorithm model, and perform conversion configuration on the algorithm model; Based on the conversion configuration result of the algorithm model, adjusting the processing state of the algorithm model by the platform end includes: Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; Parsing the extracted files to generate code data that matches the conversion of the algorithm model, and recording the code data into the algorithm model, thereby converting and configuring the algorithm model; Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

5. File capture and configuration system for algorithm model conversion, characterized in that: include: A file attribute determination module, used to obtain conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process; A file search module, used to send a file capture broadcast message to all servers connected to the platform end based on the file attribute information, so as to determine all servers that can provide the corresponding file; A file capture processing module, for capturing corresponding file data from each server end based on the file status information of each server end that can provide the corresponding file, and returning the corresponding file data packet to the platform end; A file data screening and processing module, used to identify and process file data packets received by the platform, screen file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to a queue for configuration files; An algorithm model conversion configuration module is used to extract corresponding files from the queue of configuration files based on the conversion log of the algorithm model, parse the extracted files, generate code data matching the conversion of the algorithm model, and perform conversion configuration on the algorithm model; An algorithm model processing adjustment module, used to adjust the processing status of the algorithm model by the platform end based on the conversion configuration result of the algorithm model; The file attribute determination module is used to obtain the conversion progress information of the algorithm model on the platform side, and based on the conversion progress information, predict the file attribute information required by the algorithm model during the conversion process, including: Monitor the algorithm model conversion program on the platform end to obtain conversion progress information of the algorithm model conversion task performed by the algorithm model conversion program; compare the conversion progress information with the preset algorithm model conversion calendar to predict the file type attribute information required by the corresponding algorithm model in the current conversion process; The file search module is used to send a file capture broadcast message to all servers connected to the platform end based on the file attribute information, so as to determine all servers that can provide the corresponding file, including: Based on the file type attribute information, a file capture broadcast message is sent to all active servers connected to the platform; the file directories returned by all servers regarding the file capture broadcast message are identified to determine whether the server contains the files required for the algorithm model conversion; if so, the corresponding server is determined as a server that can provide the corresponding files; if not, the corresponding server is determined as a server that cannot provide the corresponding files.

6. The file capture and configuration system for algorithm model conversion according to claim 5, characterized in that: The file capture processing module is used to capture corresponding file data from each server based on the file status information of all servers that can provide corresponding files, and return the corresponding file data packet to the platform end, including: Based on the file storage location information of all servers that can provide the corresponding files, the corresponding file data is captured from each server, and the captured file data is compressed into a file data packet and then returned to the platform end; The file data screening and processing module is used to identify and process the file data packets received by the platform end, screen the file data packets that match the conversion operation of the algorithm model, and add the screened file data packets to the queue for configuration files, including: Perform file data integrity identification processing on the file data packet received by the platform end to determine whether there is any data missing in the file data packet, and use the file data packet without data missing as the file data packet that matches the conversion operation of the algorithm model; then, based on the configuration time sequence information corresponding to the conversion operation of the file data packet that matches the conversion operation of the algorithm model, add the file data packet that matches the conversion operation of the algorithm model to the corresponding position of the queue of configuration files to be configured.

7. The file capture and configuration system for algorithm model conversion according to claim 5, characterized in that: The algorithm model conversion configuration module is used to extract corresponding files from the queue of configuration files based on the conversion log of the algorithm model, parse the extracted files, generate code data matching the conversion of the algorithm model, and perform conversion configuration on the algorithm model, including: Analyze the conversion log of the algorithm model, determine the real-time conversion process of the algorithm model, and extract the corresponding file from the queue of configuration files to be configured; parse the extracted file to generate code data that matches the conversion of the algorithm model, and record the code data in the algorithm model, so as to convert and configure the algorithm model; The algorithm model processing adjustment module is used to adjust the processing state of the algorithm model by the platform end based on the conversion configuration result of the algorithm model, including: Based on the conversion configuration result of the algorithm model, determine whether the algorithm model successfully executes the current conversion process. If so, keep the platform's conversion process of the algorithm model unchanged; if not, re-execute the platform's conversion process of the algorithm model.

Images