128 results about "Subroutine" patented technology

The embodiment of the invention provides a large language model optimization method based on variance reduction and momentum acceleration. The method comprises the following steps: a gradient estimation stage of a large language model: initializing a seed list and a projection list; executing multiple independent query iterations, calling a disturbance subprogram, generating a random seed for the large language model, and storing a table; in the disturbance subprogram, the determined gradient projection value is stored in a projection list, and iteration is carried out for next query; after executing multiple independent query iterations, storing a plurality of random seeds and a plurality of gradient projection values corresponding to the random seeds; in the weight updating stage of the large language model, a gradient norm subprogram is called for each layer of the large language model, and a random seed is obtained to reset a random number generator; and determining variance-reduced gradient estimation according to the gradient projection value extracted from the projection list and the reproduced disturbance vector. According to the embodiment of the invention, gradient information queried for multiple times is aggregated to generate low-noise gradient estimation, and fine tuning of a large language model is completed.

The invention provides a test program generation method, electronic equipment and a storage medium, relates to the field of chips, and is mainly technically characterized by comprising the following steps: generating a plurality of subprograms corresponding to each module in at least two modules according to instruction characteristics of the at least two modules in a target processor; wherein each subprogram comprises dependency information with other subprograms, each subprogram comprises a plurality of instruction fragments, each instruction fragment comprises at least one instruction, and the dependency information of the subprogram is used for indicating whether the instruction fragment in the subprogram conflicts with the instruction fragments in other subprograms or not; the to-be-used instruction fragments are continuously selected until a preset ending condition is met, and the next to-be-used instruction fragment in every two adjacent to-be-used instruction fragments is the to-be-used instruction fragment which is sequentially selected from any subprogram of any module and does not conflict with the previous to-be-used instruction fragment; and obtaining a target test program according to a selection result of the to-be-used instruction fragment.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating an output sequence from an input sequence. In one aspect, one of the systems includes one or more subroutine neural networks each comprising: an encoder neural network configured to receive an encoder network input comprising i) a subroutine input sequence of the subroutine neural network and ii) an input mask that masks one or more subroutine input elements of the subroutine input sequence and to generate an encoded representation of the encoder network input; a decoder neural network configured to receive the encoded representation and to generate the subroutine network output; and a masking neural network configured to generate an output mask, wherein the output mask will be used as the input mask of the encoder neural network at a subsequence time step of the plurality of time steps.

The invention provides an automatic testing method, device and platform for equipment load function performance, and relates to the technical field of software testing. According to the invention, through setting the plurality of subprogram packages in the process of testing the load function performance of the equipment, automatic programming is carried out according to the equipment type, the load type and the function type, the modularized combination generation of the test item, the test program and the subprogram is realized, the professional threshold is reduced, and the generation speed of the test program is improved. Furthermore, in combination with the task queue of the test accompanying instrument, path optimization is carried out on the test process, the test sequence of a plurality of test items is optimized, and the waiting time is shortened. Finally, based on the testing process and the testing program, automatic testing is achieved, the problem that the testing efficiency is low in the equipment load function performance testing process is solved, and the testing efficiency of the equipment load function performance testing is improved.

The invention belongs to the technical field of vehicle software development, and particularly relates to a method for converting a form file into a graphic program, which is used for converting the form file into the graphic program and comprises the following steps of: reading the form file so as to read text contents and coordinate information of all cells in at least one page of the form file into a corresponding two-dimensional object array, the name of each two-dimensional object array is set as a corresponding page name; analyzing each two-dimensional object array into a flow chart area and an attribute area in sequence according to pages; constructing a graphic program according to an analysis result; and / or converting the graphic program into a table file, including creating table empty pages, the number of the table empty pages being equal to the sum of the number of the main program and the number of the nested subprograms of the graphic program; corresponding table pages are sequentially constructed according to program categories in the graphic program, one page corresponds to the main program, and the other pages correspond to the corresponding nested subprograms; and storing the constructed form file.

The invention relates to a backtracking video generation method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: in response to a trigger operation of a target object for a subprogram, generating and reporting a first record identifier matched with a native page of the subprogram under the condition that the subprogram starts a backtracking function; in response to a first interaction event triggered on the native page, generating and reporting native page record data of the native page; responding to a second interaction event triggered on the webpage linked with the subprogram, and generating and reporting webpage record data of the webpage; and under the condition that a backtracking request carrying the first record identifier is received, generating a target backtracking video for the target object based on the native page record data and the web page record data which are stored in association with the first record identifier. By adopting the method, the accuracy of backtracking video generation can be improved.

The application provides a data processing method, device, equipment, medium and program product. The method comprises the following steps: obtaining a transmission instruction for transmitting subprogram data of a target subprogram of a call application; generating an establishment request of a program data channel in response to the transmission instruction and sending the establishment request to a called terminal through a control data channel, so that the called terminal detects the establishment request; if the establishment confirmation information returned based on the control data channel after the detection is passed, creating the program data channel and transmitting the subprogram data of the target subprogram based on the program data channel, thereby improving the processing efficiency of the application program data and improving the application use experience of the user.

This invention provides a task scheduling method, system, terminal device, and medium for electricity meters. The method includes: determining the type of each application of the target electricity meter; obtaining the total time slice to be allocated to all applications; wherein the application types include system applications, business applications, and container applications, and container applications include subroutines; allocating time to each application based on the total time slice according to the application priority, to obtain a first time slice for each application; allocating time to each subroutine based on the first time slice of the container application according to a preset execution time ratio, to obtain a second time slice for each subroutine; and controlling the central processing unit to execute the tasks and subroutines in each application sequentially according to the allocation results of the first and second time slices. This can improve the task processing efficiency of the electricity meter.

The invention relates to a code generation method and system based on a large language model, and belongs to the technical field of computers. According to the method, dialogue is carried out on the basis of the generated first version code and the large language model to provide more detailed and specific input, so that the large language model generates a corresponding subprogram, and complete code logic is constructed; calling an example code stored in an example code library to further optimize the initial code based on multiple rounds of conversations, and generating a second version of code; and performing comprehensive quality evaluation on the generated second version code based on a preset code quality evaluation standard, automatically generating a repair rule based on an evaluation result, and generating a target code by using the repair rule. According to the method and the device, the repair rule is obtained according to the evaluation result obtained by performing comprehensive quality evaluation on the second-version code, the second-version code is repaired based on the repair, so that the high-quality target code is obtained, and the whole generation process is more intelligent.

The invention relates to the technical field of attack tracing, in particular to an attack tracing method and device based on a symbolic execution engine, and the method comprises the steps: recognizing a plurality of dependent explosion functions of a pre-constructed execution unit sequence, so as to construct an independent subprogram; constructing a system call control flow diagram of a historical event processing cycle by combining static and dynamic analysis; auditing logs containing parameter values are collected in the execution period of the target program so as to construct a thread-level traceability graph; and according to the system call control flow diagram, determining whether a node of dependent explosion exists in the thread-level traceability diagram, if so, reconstructing the independent subprogram by using the node of dependent explosion, and analyzing the reconstructed independent subprogram by using a symbolic execution engine to reconstruct an attack path. Therefore, the problem that a related tracing attack method needs to carry out instrumentation on a system, or depends on static analysis, or depends on an unreliable application log, or needs to use additional auxiliary hardware to realize tracking is solved.

The application provides a kind of evaluation method and system of neural network cost model based on category guidance, it is related to data processing technical field, the feature of the tensor program to be evaluated is extracted from the perspective of syntax tree, and the tensor program feature is obtained;Tensor program features are processed by dimension reduction and clustering, to obtain the probability that the tensor program feature belongs to different categories;The feature of the tensor program to be evaluated is extracted from the perspective of scheduling primitive, to obtain intermediate features;Based on the probability that the tensor program feature belongs to different categories and the intermediate features, the performance score of the tensor program to be evaluated is obtained.From the perspective of syntax tree, extracting the features of the tensor program and extracting the features from the scheduling primitive can effectively utilize the complex relationship between subprograms, reduce the time of the AI compiler automatic tuning process, so that the neural network model can be deployed on the ground, and only needs less tuning time under the same tuning performance condition, improves the efficiency.

The invention discloses a deep learning library API fuzzy testing method based on semantic analysis, and belongs to the field of reliability testing of deep learning systems. The method comprises the following steps of: 1, analyzing an input check logic by using a large language model, extracting an edge condition and forming a structured expression; 2, converting the edge conditions into formalized logic predicates, performing semantic embedding and vectorization processing, and selecting representative conditions to form a variation condition set; 3, generating an executable test program, and automatically repairing a non-running program to form an executable seed program; and 4, generating diversified test input, running at the rear ends of the CPU and the GPU respectively, and identifying potential implementation defects of the deep learning framework by detecting collapse behaviors, running exceptions and cross-device output inconsistency. According to the method, the boundary condition coverage capability can be effectively improved, and the defect detection efficiency of the deep learning library API is improved.

The present invention discloses a method, device, storage medium, and processor for determining processing time. The method includes: obtaining operation information of a preset program at preset time intervals, wherein the preset program is used to control processing equipment to process a target workpiece, the target workpiece being the workpiece for which processing time is to be calculated, and the preset program includes: multiple preset subroutines, the preset subroutines including a first preset subroutine and a second preset subroutine, the first preset subroutine being used to indicate the start of the preset program, and the second preset subroutine being used to indicate the end of the preset program, and the operation information including: an acquisition time and a program identifier; identifying a first acquisition time corresponding to the first program identifier of the first preset subroutine based on the operation information; identifying a second acquisition time corresponding to the second program identifier of the second preset subroutine based on the operation information; and determining the processing time of the target workpiece based on the first acquisition time and the second acquisition time. This method solves the technical problem of the prior art in being unable to determine processing time.

To provide a game machine capable of effectively saving a use capacity regarding a specific program storage area of a main ROM.SOLUTION: In a program stored in a program area within a use area of a main ROM, an instruction size can be reduced when essentially calling a sub-program at addresses at and subsequent to a specific address by intermediating a jump instruction in consideration of a situation in which the instruction size becomes large when an instruction (for example, CALLEX) when calling a sub-program stored in a program area outside the use area calls the sub-program at addresses at and subsequent to the specific address in the program area outside the use area of the main ROM.SELECTED DRAWING: Figure 193

The embodiment of the invention provides a subprogram instance creating method and device, equipment, a medium and a program product. The method comprises the following steps: in response to a calling request of a target subprogram in a session between a client and a database, searching a target shared instance from a shared memory; multiplexing the public data in the searched target shared instance, and creating a private instance structure; and generating a private instance stored in the private memory of the session according to the private instance structure and the private data. According to the method, when the session calls the subprogram, the shared instance is searched from the shared memory, and the private instance for the session is created by multiplexing the public data in the shared instance, so that multiple sessions do not need to repeatedly compile and store the same public data, and the memory overhead is reduced, and therefore, in a high-concurrency scene, the session can be quickly and efficiently called. The risk of insufficient memory is avoided, and the stability of database operation is improved.

The present application relates to a SOC verification and optimization method, apparatus, computer equipment, and medium based on instruction stream analysis, which uniformly samples subroutines of a program and extracts Basic Block sequence information in the subroutines; extracts instruction information of each Basic Block; aggregates instruction information and Basic Block sequence information to obtain an instruction sequence, and constructs an instruction semantic sequence model based on an autoregressive language model based on the instruction sequence; constructs a Basic Block program behavior representation model based on the instruction semantic sequence model and using a comparative learning method; reads several sliced ​​subroutines to be run, and obtains the semantic representation of the sliced ​​subroutines through the Basic Block program behavior representation model; constructs a subroutine similarity matrix based on the semantic representation of the sliced ​​subroutine, and performs program selection based on the subroutine similarity matrix.

Embodiments of the disclosed technology relate to transforming high-level quantum computer programs into one or more symbolic expressions. Because the transformations result in symbolic expressions in the compiled code, one can extract these symbolic expressions to obtain a symbolic resource estimate for the quantum program. In cases where these transformations do not yield a closed-form solution, they can still be evaluated orders of magnitude faster than possibly using other resource estimation tools. Access to such symbolic or near-symbolic expressions not only greatly improves the performance of accuracy management and resource estimation, but also can better inform quantum software developers of potential bottlenecks in the quantum program. In turn, the underlying quantum computer program can be improved as appropriate.

Provided is a processor enabling assembly code of a direct-addressing conditional-jump subroutine to be reduced. The processor includes a logic circuit executing a conditional-jump instruction of assembly code, and logically configured to: store, into an index register, an index register number that is an index to a jump-destination effective address; store jump-destination effective addresses differing from one another, or address offsets to jump-destination effective addresses differing from one another, into plural reference registers, respectively; fetch and decode a conditional jump instruction that is one instruction that quotes the index register; and then execute (S1) acquiring a jump-destination effective address or an address offset from the reference register corresponding to the index register number of the content of the index register and (S2) setting the acquired jump-destination effective address, or a jump-destination effective address calculated from the address offset, to a register serving as the program counter.

A method implements generating a sampling profile and an instrumentation profile from source code. The method involves identifying a hot subroutine from the source code with the sampling profile. The method further involves executing a hot modifier process with the hot subroutine and one or more of the sampling profile and the instrumentation profile to generate a hot compilation unit. The method further involves executing a cold modifier process with a cold subroutine from the source code and the instrumentation profile to generate a cold compilation unit. The method further involves incorporating the hot compilation unit and the cold compilation unit into a modified compilation schedule.

The application discloses a PLC program reverse analysis method and device. The method comprises the following steps: obtaining a to-be-processed file; performing structural analysis processing on the to-be-processed file to obtain program entry address information and memory initialization information; performing subprogram identification processing on the to-be-processed file according to the program entry address information and the memory initialization information to obtain program flow feature data; performing identification processing on the program flow feature data based on a library function signature to obtain standard function block data; performing parameter extraction processing on the program flow feature data to obtain control behavior parameter data; and performing visual modeling processing on the program entry address information, the memory initialization information, the program flow feature data, the standard function block data and the control behavior parameter data to obtain reverse analysis result data. The automatic and structured control program reverse reconstruction capability is realized.

This specification provides a method and apparatus for processing rights and interests. One method includes: obtaining the code value of a store service code uploaded by a program container deployed by an application carrying a subroutine; the code value is obtained by the subroutine collecting and decoding the store service code; querying a store identifier mapped to the code value in a code value-store identifier mapping table; determining the rights and interests information of the target store corresponding to the store identifier and sending it to the program container, so that the program container can send the rights and interests information to the subroutine for display.

The present disclosure provides a method and device for restoring obfuscation of an Android string, a computer device and a medium, comprising: converting intermediate bytecode of an Android application installation package data into a first representation file and a second representation file; performing semantic understanding on the second representation file to obtain a control flow graph corresponding to a target method body; identifying obfuscation points based on the control flow graph corresponding to the target method body to obtain a plurality of target candidate sentences; generating a corresponding slice standard representation for each target candidate sentence; reversely constructing a program dependency chain based on the plurality of target candidate sentences to obtain a data flow graph structure; determining a minimum code subset based on the data flow graph structure and the slice standard representation corresponding to each target candidate sentence; performing a reorganization operation on the minimum code subset to generate a subprogram; embedding the subprogram into a class file to obtain a target class file; and executing the target class file through the first representation file to obtain a decrypted string. Thus, the success rate of string decryption is improved.

The embodiment of the present specification provides a prompt sound test method and device of a vehicle, wherein a prompt sound test method of a vehicle comprises: obtaining a test instruction of a prompt sound corresponding to prompt sound configuration data set for the vehicle through a subprogram; the subprogram comprises a first subprogram in an application program running on a user terminal; loading the prompt sound configuration data according to the response state of the sound playing component of the vehicle, and playing the prompt sound after loading; judging whether the collected test prompt sound meets the prompt sound index corresponding to the environmental sound data; if yes, it is determined that the test of the prompt sound is passed.

This application proposes a laser processing method, apparatus, processing equipment, and storage medium for circuit boards, relating to the field of laser processing technology. The method obtains the processing requirements of the circuit board; these requirements include processing sub-requirements for multiple processing structures within the circuit board; different processing subroutines are then set based on these sub-requirements, forming a processing program; and laser processing is then performed on multiple processing structures according to this processing program. This application customizes processing subroutines for each processing structure based on its specific processing sub-requirements, ensuring that the final processing program includes subroutines that meet different processing sub-requirements. When performing laser processing based on this program, laser processing can be carried out according to the processing sub-requirements of each processing substructure. Only one processing drawing is needed, and a complete processing program can be executed to process the entire circuit board, resulting in higher processing efficiency.

The invention discloses a numerical control machining tool checking method and system and a medium, and relates to the technical field of numerical control machining.The numerical control machining tool checking method comprises the following steps that the structural size of a tool checking block is obtained, a tool checking block model consistent with the structural size of the tool checking block is established, and the tool checking block is a conical tool checking block; a part machining main program is obtained, and a machining main program preposed file is converted into a G code of an NC program through postprocessing; current tool parameters are obtained, the current tool parameters are transmitted to a packaged fixed loop program, and the fixed loop program is used for calling a parameterized preset tool checking block subprogram; and executing an NC program including the fixed loop program, suspending processing when the NC program runs to a selective suspending instruction, and performing integrity check on the tool checking block. Dependence on experience of operators is reduced, human intervention risks are reduced, and machining automation and tool checking efficiency are improved.

The invention belongs to the technical field of vehicle software development, and particularly relates to a table file and graphic program conversion system, which comprises a computer device, and the computer device is configured to comprise a table and graphic conversion module which is configured to read a table file and read a graphic program; reading the text content and coordinate information of all cells in at least one page of the table file into the corresponding two-dimensional object arrays, and setting the name of each two-dimensional object array as the name of the corresponding page; analyzing each two-dimensional object array into a flow chart area and an attribute area in sequence according to pages; constructing a graphic program according to an analysis result; and / or the graph and table conversion module is configured to create table empty pages, and the number of the table empty pages is equal to the sum of the number of main programs and the number of nested subprograms in the graph program; corresponding table pages are sequentially constructed according to program categories in the graphic program, one page corresponds to the main program, and the other pages correspond to the corresponding nested subprograms; and storing the constructed form file.

The present invention relates to a method for setting a workflow of a chromatography system, the workflow comprising a plurality of workflow parameters, the method comprising: providing a subroutine of the workflow; providing at least one boundary condition for at least one workflow parameter; assigning a duration and a start time to each of the sub-routines; and generating the workflow by combining the subroutines.

The application provides a model training method, which comprises the following steps: determining a sample abnormal subprogram in a sample program, and determining sample syntax structure information of the sample abnormal subprogram based on the sample abnormal subprogram; constructing an initial program repair model, and setting state information, action information and reward information for the initial program repair model; training the initial program repair model after the setting based on the sample abnormal subprogram and the sample syntax structure information by using a target reinforcement learning algorithm, to obtain a target program repair model, thereby solving the problems of poor efficiency and low accuracy of the related art in repairing the smart contract. The application also discloses a program repair method, an electronic device, a storage medium and a program product.