Execution method and device of model reasoning task, equipment and storage medium

Abstract

Embodiments of the invention provide a model reasoning task execution method and apparatus, a device and a storage medium. The method comprises the steps of obtaining a model reasoning task initiated by a user; determining a computing node for executing the model reasoning task; sending task data corresponding to the model reasoning task to a computing node, and controlling the computing node to execute the model reasoning task according to the task data to obtain a model reasoning result corresponding to the computing node; counting the number of model reasoning results corresponding to the computational nodes to obtain a target number corresponding to the computational nodes; determining target computing nodes according to the target number corresponding to the computing nodes; and controlling the target computing node to generate a target block in the block chain, and storing a model reasoning result corresponding to the target computing node in the target block. According to the embodiment of the invention, the model reasoning task is allocated to the computing node, so that the computing node can execute the valuable computing task, the utilization efficiency of computing resources is improved, and the waste of the resources is avoided.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and device for executing a model reasoning task, an electronic device, and a storage medium. Background Art

[0002] Blockchain technology is a distributed ledger system that maintains a growing record of transaction data in a decentralized manner. The data information of each newly generated block contains a summary of the content of the previous block, ensuring that the data added to the chain cannot be tampered with. The blockchain system needs to use a consensus mechanism to decide which node adds which data to the blockchain. Some public blockchains solve a mathematical problem so that the first node to complete the calculation will add the data to the blockchain. Since the solution process requires a lot of computing resources, this consensus mechanism is also called Proof of Work (PoW).

[0003] However, the PoW solution process does not produce any valuable computing results except for selecting nodes for generating new blocks, which results in a large waste of computing resources.

[0004] Therefore, how to make full use of the computing resources consumed in the PoW process is an issue that needs to be urgently addressed in this field. Summary of the invention

[0005] The embodiment of the present application provides a method for executing a model reasoning task to solve the problem that the PoW solution process causes a large amount of computing resources to be wasted.

[0006] Correspondingly, an embodiment of the present application also provides a device for executing a model reasoning task, an electronic device and a storage medium to ensure the implementation and application of the above method.

[0007] In order to solve the above problems, the embodiment of the present application discloses a method for executing a model reasoning task, which is applied to a blockchain system, wherein the blockchain system includes a blockchain and a user, and the method includes:

[0008] Acquire a model reasoning task initiated by the user; the model reasoning task has corresponding task data;

[0009] Determining a computing node for executing the model reasoning task;

[0010] Sending the task data corresponding to the model reasoning task to the computing node, and controlling the computing node to perform the model reasoning task according to the task data, to obtain the model reasoning result corresponding to the computing node;

[0011] Counting the number of model inference results corresponding to the computing nodes to obtain the target number corresponding to the computing nodes;

[0012] Determine the target computing node according to the target number corresponding to the computing node;

[0013] The target computing node is controlled to generate a target block in the blockchain, and the model inference result corresponding to the target computing node is stored in the target block.

[0014] Optionally, the blockchain system further includes an arbitration node. After sending the task data corresponding to the model reasoning task to the computing node and controlling the computing node to perform the model reasoning task according to the task data, and obtaining the model reasoning result corresponding to the computing node, the method further includes:

[0015] Controlling the computing node to send the model reasoning result to the arbitration node;

[0016] Determining a verification node for verifying the model reasoning result;

[0017] Sending the task data corresponding to the model reasoning task to the verification node, and controlling the verification node to verify the model reasoning result according to the task data to obtain a model verification result;

[0018] Controlling the verification node to send the model verification result to the arbitration node;

[0019] The arbitration node is controlled to compare whether the model inference result and the model verification result are the same, and obtain an arbitration result for determining whether the computing node is a malicious node; the malicious node is a computing node or a verification node that affects the normal operation of the blockchain system.

[0020] Optionally, the blockchain system is in communication with a model reasoning task provider, the model reasoning task provider includes a standard reasoning result corresponding to the model reasoning task, and the controlling the arbitration node compares whether the model reasoning result and the model verification result are the same, and obtains an arbitration result for determining whether the computing node is a malicious node, including:

[0021] If the model reasoning result and the model verification result are the same, a first arbitration result for determining that the computing node is not the malicious node is obtained;

[0022] If the model reasoning result and the model verification result are different, obtaining the standard reasoning result corresponding to the model reasoning task from the model reasoning task provider, and obtaining a second arbitration result for determining whether the computing node or the verification node is the malicious node based on the model reasoning result, the model verification result and the standard reasoning result;

[0023] The first arbitration result or the second arbitration result is determined as the arbitration result.

[0024] Optionally, obtaining, according to the model reasoning result, the model verification result and the standard reasoning result, a second arbitration result for determining whether the computing node or the verification node is the malicious node includes:

[0025] comparing whether the model reasoning result is the same as the standard reasoning result, and comparing whether the model verification result is the same as the standard reasoning result;

[0026] If the model reasoning result is different from the standard reasoning result, a third arbitration result for determining that the computing node is the malicious node is obtained;

[0027] If the model reasoning result is the same as the standard reasoning result, a fourth arbitration result for determining that the computing node is not the malicious node is obtained;

[0028] If the model verification result is different from the standard reasoning result, obtaining a fifth arbitration result for determining that the verification node is the malicious node;

[0029] If the model verification result is the same as the standard reasoning result, a sixth arbitration result for determining that the verification node is not the malicious node is obtained;

[0030] The third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result are determined as the second arbitration result.

[0031] Optionally, the computing node has a corresponding number of unresponsive times, and sending the task data corresponding to the model reasoning task to the computing node, and controlling the computing node to perform the model reasoning task according to the task data, to obtain the model reasoning result corresponding to the computing node, includes:

[0032] Sending the task data corresponding to the model reasoning task to the computing node;

[0033] Determining whether the computing node responds to the model reasoning task within a preset time;

[0034] If the computing node does not respond to the model inference task within a preset time, the number of non-response times corresponding to the computing node is increased by one.

[0035] Optionally, the computing node has computing power for executing the model reasoning task, and after the number of non-response times corresponding to the computing node is increased by one if the computing node does not respond to the model reasoning task within a preset time, the method further includes:

[0036] If the number of non-response times corresponding to the computing node exceeds a preset number, the computing capacity of the user corresponding to the computing node is cancelled.

[0037] Optionally, the computing node has computing capability for executing the model reasoning task, and the verification node has verification capability for verifying the model reasoning result. After determining the third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result as the second arbitration result, the method further includes:

[0038] If the computing node is the malicious node, canceling the computing capability of the computing node;

[0039] If the verification node is the malicious node, the verification capability of the verification node is cancelled.

[0040] The embodiment of the present application also discloses a device for executing a model reasoning task, which is applied to a blockchain system, wherein the blockchain system includes a blockchain and a user, and the device includes:

[0041] A model reasoning task acquisition module, used to acquire the model reasoning task initiated by the user; the model reasoning task has corresponding task data;

[0042] A computing node determination module, used to determine a computing node for executing the model reasoning task;

[0043] A model reasoning task execution module, used to send the task data corresponding to the model reasoning task to the computing node, and control the computing node to execute the model reasoning task according to the task data, so as to obtain the model reasoning result corresponding to the computing node;

[0044] A target quantity statistics module is used to count the number of model inference results corresponding to the computing nodes to obtain the target quantity corresponding to the computing nodes;

[0045] A target computing node determination module, used to determine the target computing node according to the target number corresponding to the computing node;

[0046] The target block generation module is used to control the target computing node to generate a target block in the blockchain, and store the model inference result corresponding to the target computing node in the target block.

[0047] An embodiment of the present application also discloses an electronic device, including: a processor; and a memory, on which executable code is stored. When the executable code is executed, the processor executes a method for executing a model reasoning task as described in any one of the embodiments of the present application.

[0048] The embodiments of the present application also disclose one or more machine-readable media on which executable codes are stored. When the executable codes are executed, the processor executes the execution method of the model reasoning task as described in any one of the embodiments of the present application.

[0049] Compared with the prior art, the embodiments of the present application have the following advantages:

[0050] In an embodiment of the present application, a model reasoning task initiated by the user is obtained; the model reasoning task has corresponding task data; a computing node for executing the model reasoning task is determined; the task data corresponding to the model reasoning task is sent to the computing node, and the computing node is controlled to execute the model reasoning task according to the task data, and the model reasoning result corresponding to the computing node is obtained; the number of model reasoning results corresponding to the computing node is counted to obtain the target number corresponding to the computing node; the target computing node is determined according to the target number corresponding to the computing node; the target computing node is controlled to generate a target block in the blockchain, and the model reasoning result corresponding to the target computing node is stored in the target block. The embodiment of the present application replaces mathematical problems with model reasoning tasks, and by assigning model reasoning tasks to computing nodes, computing nodes are able to perform valuable computing tasks, improve the utilization efficiency of computing resources, and avoid waste of resources. In addition, the selected target computing node is responsible for generating the target block, which ensures the fairness and transparency of the block generation process, further enhances the credibility of blockchain data, and stores the model reasoning results in the target block of the blockchain. Due to the characteristics of the blockchain, these data cannot be tampered with once recorded, ensuring the credibility and security of the model reasoning results. In addition, the target computing node is determined according to the number of model inference results corresponding to the computing node, avoiding the situation where a single computing node monopolizes computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1It is a flowchart of the steps of an embodiment of a method for executing a model reasoning task of the present application;

[0052] Figure 2 It is a schematic diagram of a target computing node generating a target block in the present application;

[0053] Figure 3 It is a flowchart of an arbitration node in the present application recording user capabilities in a blockchain system;

[0054] Figure 4 This is a flowchart of a user initiating a model reasoning task in a blockchain system in this application;

[0055] Figure 5 This is a flow chart of a model-based reasoning task provider arbitrating nodes in this application;

[0056] Figure 6 It is a structural block diagram of an embodiment of an execution device of a model reasoning task of the present application;

[0057] Figure 7 It is a schematic diagram of the structure of a device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0059] Reference Figure 1 , is a flowchart of a method for executing a model reasoning task in the present application, which is applied to a blockchain system, wherein the blockchain system includes a blockchain and a user, and includes the following steps:

[0060] Step 101: obtaining a model reasoning task initiated by the user; the model reasoning task has corresponding task data;

[0061] In the embodiment of the present application, the user is the basic role in the blockchain system.

[0062] First, the model inference task initiated by the user will be received. The user submits a task request through an interface or protocol in the blockchain system. The task request contains specific task data, which is the input information required for model inference.

[0063] Step 102: determining a computing node for executing the model reasoning task;

[0064] In the embodiment of the present application, each user will have corresponding computing resources. If the user has computing power and sufficient computing resources, he can participate in the blockchain system as a computing node and a user. While contributing his own computing power, he can also use the large model capabilities provided by other computing nodes to realize computing power sharing.

[0065] Specifically, one or more computing nodes will be selected to execute the model inference tasks submitted by the user. These computing nodes are users with computing capabilities and abundant computing resources in the blockchain system, but they are not the users who submitted the model inference tasks.

[0066] Step 103: sending the task data corresponding to the model reasoning task to the computing node, and controlling the computing node to perform the model reasoning task according to the task data, to obtain the model reasoning result corresponding to the computing node;

[0067] After determining the computing node for executing the model reasoning task, the system will send the task data corresponding to the model reasoning task to the selected computing node and control these computing nodes to execute the model reasoning task according to the received task data. After the computing node completes the reasoning, it will generate the corresponding model reasoning result.

[0068] The computing nodes are able to respond and execute tasks quickly and perform reasoning based on task data to ensure the efficient completion of reasoning tasks and the accuracy and reliability of reasoning results.

[0069] In the embodiment of the present application, the computing node can receive and process the model reasoning task initiated by the user, and the corresponding token and score can be obtained by completing the model reasoning task. The computing node needs to ensure that it is online and in a processing listening state to process the model reasoning task initiated by the user in a timely manner.

[0070] Step 104: Count the number of model inference results corresponding to the computing nodes to obtain the target number corresponding to the computing nodes;

[0071] The system will count the number of model inference tasks completed by each computing node within a certain period of time and obtain the target number of each computing node. This number directly reflects the activity and contribution of the computing node.

[0072] The embodiment of the present application can quantify the contribution of each computing node to the execution of the model reasoning task by counting the number of inference results, so as to encourage more users to participate in the execution of the model reasoning task as computing nodes.

[0073] Step 105: determining a target computing node according to the target number corresponding to the computing node;

[0074] Select one or more target computing nodes based on the target number of computing nodes. These nodes are the nodes that complete the number of tasks that meet the preset requirements.

[0075] The embodiments of the present application ensure the fairness and rationality of the selection of target computing nodes by selecting a target number, thereby avoiding a single computing node from monopolizing the resources of the blockchain system. In addition, the selection of target computing nodes can effectively motivate more users to participate in the execution of model reasoning tasks as computing nodes, thereby improving the activity of the blockchain system.

[0076] Step 106: Control the target computing node to generate a target block in the blockchain, and store the model inference result corresponding to the target computing node in the target block.

[0077] The system will control the target computing node to generate a new block in the blockchain, and store all model inference results of the target computing node in this new block. These results will be permanently recorded on the blockchain and cannot be tampered with.

[0078] Reference Figure 2 , is a schematic diagram of a target computing node generating a target block in the present application.

[0079] In the blockchain system, each block has a unique number, such as Figure 2 Block10, Block11, Block12, Block23, Block24 in the block chain represent the position of the block in the blockchain. Similarly, each computing node has a unique number, such as Figure 2 N-2, N, N+1, 2N-3, 2N.

[0080] Assume that the preset requirement for determining the target computing node is to execute the model reasoning task five times. Then, Figure 2 As shown, assuming that the target number of computing nodes numbered N before generating Block11 is 10, it means that the computing node numbered N has completed 10 model reasoning tasks before generating Block11. At this time, the target number of computing nodes numbered N meets the preset requirements, and it can be used as the target computing node to generate the target block Block11 in the blockchain.

[0081] The embodiment of the present application generates new blocks through the target computing node, ensuring the fairness and transparency of the block generation process. In addition, the model reasoning results are stored in the blockchain, ensuring the security and non-tamperability of the data, and enhancing the credibility of the results.

[0082] The embodiment of the present application replaces mathematical problems with model reasoning tasks. By assigning model reasoning tasks to computing nodes, computing nodes can perform valuable computing tasks, improve the utilization efficiency of computing resources, and avoid waste of resources. In addition, the selected target computing node is responsible for generating the target block, which ensures the fairness and transparency of the block generation process, further enhances the credibility of blockchain data, and stores the model reasoning results in the target block of the blockchain. Due to the characteristics of the blockchain, these data cannot be tampered with once recorded, ensuring the credibility and security of the model reasoning results. In addition, the target computing node is determined according to the number of model reasoning results corresponding to the computing node, avoiding the situation where a single computing node monopolizes computing resources.

[0083] In one embodiment of the present application, the blockchain system further includes an arbitration node. After sending the task data corresponding to the model reasoning task to the computing node and controlling the computing node to perform the model reasoning task according to the task data, and obtaining the model reasoning result corresponding to the computing node, the method further includes:

[0084] Controlling the computing node to send the model reasoning result to the arbitration node;

[0085] Determining a verification node for verifying the model reasoning result;

[0086] Sending the task data corresponding to the model reasoning task to the verification node, and controlling the verification node to verify the model reasoning result according to the task data to obtain a model verification result;

[0087] Controlling the verification node to send the model verification result to the arbitration node;

[0088] The arbitration node is controlled to compare whether the model inference result and the model verification result are the same, and obtain an arbitration result for determining whether the computing node is a malicious node; the malicious node is a computing node or a verification node that affects the normal operation of the blockchain system.

[0089] In the embodiment of the present application, in order to ensure the verifiability of the model reasoning results, in addition to the users and the blockchain, the blockchain system also has an arbitration node, which participates in the model reasoning task as a trusted third party. In the blockchain system, the arbitration node cannot actively initiate the model reasoning task, and all operations will be automatically executed according to the pre-set rules.

[0090] After completing the model inference task, the computing node will send the model inference result to the arbitration node. As a trusted third party, the arbitration node is responsible for receiving and storing the inference results. Receiving the inference results through the arbitration node ensures the credibility and security of the results and prevents malicious nodes from tampering with the model inference results.

[0091] When the arbitration node receives the model inference result, it sends the model inference result to the user who initiated the model inference task.

[0092] In an embodiment of the present application, if the user has verification capabilities and sufficient computing resources, he or she can participate in the blockchain system as both a verification node and a user.

[0093] Specifically, one or more verification nodes will be selected to verify the model inference results obtained by the computing nodes. These verification nodes are users in the blockchain system with verification capabilities and abundant computing resources, but they are not users who submit model inference tasks. By verifying the nodes, the accuracy and reliability of the model inference results can be ensured.

[0094] The verification node uses the same input data as the computing node, and judges whether the results obtained by the computing node are authentic and reliable through the output results. All verification nodes in the blockchain system can verify the model reasoning tasks performed by any computing node. Here, the verification of the model reasoning results by the verification node is called the model verification task. The specific system sends the task data of the model reasoning task to the selected verification nodes, and controls these nodes to perform the model reasoning task according to the received task data to obtain the model verification results. The verification node can respond quickly and use the same task data to perform the model reasoning task, ensuring the consistency and reliability of the model verification results.

[0095] Similarly, a verification node can also verify the model verification tasks initiated by other verification nodes. For the same model reasoning task, the results of all verification nodes participating in the verification must be completely consistent before the computing node verification can be considered successful.

[0096] After obtaining the model verification result, the verification node sends the model verification result to the arbitration node. The arbitration node is responsible for receiving and storing the model verification result. The arbitration node compares the model reasoning result of the computing node with the model verification result of the verification node to determine whether the two are consistent. If they are consistent, the model reasoning result is considered to be credible; if they are inconsistent, the computing node or the verification node is considered to have malicious behavior and generates an arbitration result.

[0097] The embodiment of the present application ensures the accuracy and credibility of the reasoning results by comparing arbitration nodes, avoiding interference from malicious nodes, and ensures the normal operation of the blockchain system and the security of data by identifying and processing malicious nodes.

[0098] Reference Figure 3 , is a flowchart of an arbitration node in the present application recording user capabilities in a blockchain system. Specifically, the arbitration node will accept each user ( Figure 2 The record operation sent by user 1 and user 2 in the block chain records the computing power of user 1 and the verification power of user 2 in the block chain.

[0099] Reference Figure 4 , is a flowchart of a user initiating a model reasoning task in a blockchain system in this application, which specifically includes the following steps:

[0100] Step 401: The user initiates a model reasoning task T infer ,The arbitration node receives the user's model inference task and stores the original data and summary.

[0101] Step 402: The computing node performs the model reasoning task and generates the model reasoning result Result Tinfer , and the model inference results Result Tinfer Perform MD5 hash calculation to generate hash value 1234..., then send hash value 1234... to the arbitration node, which receives the model inference result of the computing node and stores the inference result and summary.

[0102] Step 403: The model inference result Result Tinfer Send to user.

[0103] Step 404: Verification nodes 1, 2, and 3 verify the model inference results, and the arbitration node receives the model verification task T from the verification node. verify , generate model verification results Result Tverify , and verify the model results Result Tverify An MD5 hash calculation is performed to generate a hash value 5678..., and then the hash value 5678... is sent to the arbitration node, and the arbitration node verifies the model verification result.

[0104] Step 405: Model inference result Result Tinfer Corresponding hash value and model verification result Result Tverify If the corresponding hash values ​​are inconsistent, the computing node is marked as a malicious computing node.

[0105] The embodiment of the present application further enhances the credibility of the model reasoning results and the security of the blockchain system by introducing arbitration nodes and verification nodes. Specifically, the accuracy and credibility of the model reasoning results are ensured through multi-party verification and comparison of arbitration nodes, and the normal operation of the blockchain system and the security of data are ensured by identifying and processing malicious nodes.

[0106] In one embodiment of the present application, the blockchain system is in communication with a model reasoning task provider, the model reasoning task provider includes a standard reasoning result corresponding to the model reasoning task, and the arbitration node is controlled to compare whether the model reasoning result and the model verification result are the same, and obtain an arbitration result for determining whether the computing node is a malicious node, including:

[0107] If the model reasoning result and the model verification result are the same, a first arbitration result for determining that the computing node is not the malicious node is obtained;

[0108] If the model reasoning result and the model verification result are different, obtaining the standard reasoning result corresponding to the model reasoning task from the model reasoning task provider, and obtaining a second arbitration result for determining whether the computing node or the verification node is the malicious node based on the model reasoning result, the model verification result and the standard reasoning result;

[0109] The first arbitration result or the second arbitration result is determined as the arbitration result.

[0110] In the embodiment of the present application, the model reasoning task provider is a trusted third party responsible for providing standard reasoning results. When the model reasoning result and the model verification result are inconsistent, it is considered that the computing node or the verification node has malicious behavior. At this time, it is necessary to determine whether the computing node has malicious behavior or the verification node has malicious behavior, or whether both the computing node and the verification node have malicious behavior based on the standard reasoning result provided by the model reasoning task provider.

[0111] Specifically, if the model inference result generated by the computing node is the same as the model verification result generated by the verification node, it means that the model inference result is credible and no abnormality occurs. In this case, a first arbitration result is generated to determine that the computing node is not a malicious node, which means that the computing node is an honest node without malicious behavior.

[0112] If the model inference result generated by the computing node is different from the model verification result generated by the verification node, it means that the inference result may be abnormal and needs further verification. In this case, the standard inference result corresponding to the model inference task is obtained from the model inference task provider. Since the standard inference result is generated by a trusted third party and has a high degree of credibility, it can be determined whether the computing node and the verification node are malicious nodes with malicious behavior based on the standard inference result.

[0113] Finally, according to the results of the above comparison and verification, the first arbitration result or the second arbitration result is determined as the final arbitration result, and this arbitration result is used to determine whether the computing node or the verification node is a malicious node, and perform corresponding processing.

[0114] Reference Figure 5 , is a flow chart of a model-based reasoning task provider arbitrating nodes in this application, which specifically includes the following steps:

[0115] Step 501: The user initiates a model reasoning task T infer ,The arbitration node receives the user's model inference task and stores the original data and summary.

[0116] Step 502: The computing node performs the model reasoning task and generates the model reasoning result Result Tinfer , and the model inference results Result Tinfer Perform MD5 hash calculation to generate a hash value 5678..., and send the hash value 5678... to the arbitration node.

[0117] Step 503: The model inference result Result Tinfer Send to user.

[0118] Step 504: Verification nodes 1, 2, 3, 4 and 5 will verify the model inference results, and the arbitration node receives the model verification task T from the verification node. verify , generate model verification results Result Tverify , and verify the model results Result Tverify Perform MD5 hash calculation to generate hash values ​​(the hash values ​​corresponding to verification nodes 1, 2, and 3 are 1234..., and the hash values ​​corresponding to verification nodes 4 and 5 are 5678...), and send the hash values ​​to the arbitration node.

[0119] Step 505: Since the hash values ​​between the verification nodes are inconsistent, the arbitration node initiates an acquisition request to the trusted model reasoning task provider, requesting the standard reasoning result Result.

[0120] Step 506: The model reasoning task provider generates a standard reasoning result Result, and performs MD5 hash calculation on the reasoning result to generate a hash value 5678..., and sends the hash value to the arbitration node.

[0121] Step 507: Since the hash values ​​corresponding to verification node 1, verification node 2 and verification node 3 are inconsistent with the hash value corresponding to the standard reasoning result Result, verification node 1, verification node 2 and verification node 3 are marked as malicious node 1, malicious node 2 and malicious node 3.

[0122] It should be noted that Figure 5 The honest nodes 1 and 2 in the above-mentioned step 504 refer to the verification nodes 4 and 5. Figure 5 The honest computing node in refers to the computing node in the above step 502.

[0123] The embodiment of the present application can ensure the accuracy and credibility of the model reasoning results through this arbitration mechanism. Specifically, the accuracy and credibility of the reasoning results are ensured through multi-party verification and comparison of arbitration nodes. By identifying and processing malicious nodes, the normal operation of the blockchain system and the security of data are ensured. By introducing a trusted large model service provider and standard reasoning results, the accuracy and credibility of the arbitration results are further enhanced, ensuring the fairness and transparency of the system.

[0124] In one embodiment of the present application, obtaining, according to the model reasoning result, the model verification result and the standard reasoning result, a second arbitration result for determining whether the computing node or the verification node is the malicious node includes:

[0125] comparing whether the model reasoning result is the same as the standard reasoning result, and comparing whether the model verification result is the same as the standard reasoning result;

[0126] If the model reasoning result is different from the standard reasoning result, a third arbitration result for determining that the computing node is the malicious node is obtained;

[0127] If the model reasoning result is the same as the standard reasoning result, a fourth arbitration result for determining that the computing node is not the malicious node is obtained;

[0128] If the model verification result is different from the standard reasoning result, obtaining a fifth arbitration result for determining that the verification node is the malicious node;

[0129] If the model verification result is the same as the standard reasoning result, a sixth arbitration result for determining that the verification node is not the malicious node is obtained;

[0130] The third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result are determined as the second arbitration result.

[0131] In the model reasoning task, it is necessary to determine whether the computing node or the verification node is a malicious node by comparing the model reasoning results, the model verification results and the standard reasoning results, and finally generate the second arbitration result.

[0132] Specifically, the model reasoning results are compared to see if they are the same as the standard reasoning results, and the model verification results are compared to see if they are the same as the standard reasoning results.

[0133] If the model reasoning result is different from the standard reasoning result, it means that the reasoning result generated by the computing node is abnormal. In this case, a third arbitration result is generated to determine that the computing node is a malicious node, which means that the computing node is considered to be a malicious node with abnormal behavior.

[0134] If the model reasoning result is the same as the standard reasoning result, it means that the reasoning result generated by the computing node is correct. In this case, a fourth arbitration result is generated to determine that the computing node is not a malicious node, which means that the computing node is considered to be an honest node without malicious behavior.

[0135] If the model verification result is different from the standard reasoning result, it means that the verification result generated by the verification node is abnormal. In this case, a fifth arbitration result is generated to determine that the verification node is a malicious node, which means that the computing node is considered to be a malicious node with abnormal behavior.

[0136] If the model verification result is the same as the standard reasoning result, it means that the verification result generated by the verification node is correct. In this case, a sixth arbitration result is generated to determine that the verification node is not a malicious node, which means that the computing node is considered to be an honest node without malicious behavior.

[0137] Finally, based on the results of the above comparison and verification, one of the following combinations is determined as the second arbitration result: the third arbitration result and the fifth arbitration result (both the computing node and the verification node are malicious nodes); the third arbitration result and the sixth arbitration result (the computing node is a malicious node, and the verification node is an honest node); the fourth arbitration result and the fifth arbitration result (the computing node is an honest node, and the verification node is a malicious node); the fourth arbitration result and the sixth arbitration result (both the computing node and the verification node are honest nodes).

[0138] The embodiment of the present application ensures the accuracy and credibility of the model reasoning results through multi-party verification and comparison of arbitration nodes. By identifying and processing malicious nodes, the normal operation of the blockchain system and the security of data are ensured. By introducing a trusted large model service provider and standard reasoning results, the accuracy and credibility of the arbitration results are further enhanced, ensuring the fairness and transparency of the system. By comparing the model reasoning results, model verification results and standard reasoning results, the comprehensiveness and reliability of the arbitration results are ensured, avoiding the limitations of single verification.

[0139] In one embodiment of the present application, the computing node has a corresponding number of unresponsive times, and the task data corresponding to the model reasoning task is sent to the computing node, and the computing node is controlled to perform the model reasoning task according to the task data to obtain the model reasoning result corresponding to the computing node, including:

[0140] Sending the task data corresponding to the model reasoning task to the computing node;

[0141] Determining whether the computing node responds to the model reasoning task within a preset time;

[0142] If the computing node does not respond to the model inference task within a preset time, the number of non-response times corresponding to the computing node is increased by one.

[0143] After the system sends the task data corresponding to the model inference task initiated by the user to the selected computing node, it will check whether the computing node responds and starts executing the model inference task within a preset time range (such as a few seconds or minutes). The preset time is to ensure that the computing node can process the task in a timely manner.

[0144] If a computing node does not respond to a model inference task within the preset time, the system will increase the number of unresponsiveness of the computing node by 1. The number of unresponsiveness is a counter that records the number of times a computing node fails to respond to a task in a timely manner.

[0145] The embodiment of the present application ensures that the computing node can respond and execute the model reasoning task in time by presetting the time limit, which can avoid long waiting time for the model reasoning task. By recording the number of unresponsiveness of the computing node, the system can monitor the reliability of the computing node. By recording the number of unresponsiveness, the system can adjust the nodes that frequently fail to respond, encourage the nodes to maintain a good response state, and improve the efficiency of the overall system.

[0146] In one embodiment of the present application, the computing node has computing power for executing the model reasoning task. If the computing node does not respond to the model reasoning task within a preset time, after the number of non-response times corresponding to the computing node is increased by one, the method further includes:

[0147] If the number of non-response times corresponding to the computing node exceeds a preset number, the computing capacity of the user corresponding to the computing node is cancelled.

[0148] The computing nodes have the computing power required to perform model reasoning tasks and can respond to and process model reasoning tasks initiated by users.

[0149] In the above steps, if a computing node does not respond to the model inference task within the preset time, the system will increase the number of non-response times of the computing node by one.

[0150] If the number of unresponsive times of a computing node exceeds the preset number (such as 3 or 5 times), the system will cancel the computing power of the user corresponding to the computing node, which means that the user will no longer be able to participate in model inference tasks as a computing node.

[0151] The embodiment of the present application ensures that the computing nodes can respond in time and execute model reasoning tasks by presetting time limits and monitoring the number of unresponsive times, thereby avoiding long waiting times for tasks. By recording and processing the number of unresponsive times, the system can monitor the reliability of the computing nodes. By recording the number of unresponsive times and canceling computing power, the system can adjust nodes that frequently fail to respond, incentivize nodes to maintain a good response state, and improve the efficiency of the overall system. By canceling computing power, the system can regulate user behavior, avoid malicious users or unreliable users from abusing computing resources, and ensure the fairness and security of the system.

[0152] In one embodiment of the present application, the computing node has a computing capability for executing the model reasoning task, and the verification node has a verification capability for verifying the model reasoning result. After determining the third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result as the second arbitration result, the method further includes:

[0153] If the computing node is the malicious node, canceling the computing capability of the computing node;

[0154] If the verification node is the malicious node, the verification capability of the verification node is cancelled.

[0155] The verification node has the verification capabilities required to verify the model reasoning results and can respond to and verify the reasoning results of the computing node.

[0156] In the above steps, the system generates a second arbitration result by comparing the model reasoning result, the model verification result and the standard reasoning result. Specifically, the third arbitration result is that the computing node is a malicious node; the fourth arbitration result is that the computing node is not a malicious node; the fifth arbitration result is that the verification node is a malicious node; the sixth arbitration result is that the verification node is not a malicious node.

[0157] If the second arbitration result determines that the computing node is a malicious node, the system will cancel the computing power of the computing node, which means that the node will no longer be able to participate in the model reasoning task as a computing node.

[0158] If the second arbitration result determines that the verification node is a malicious node, the system will cancel the verification capability of the verification node, which means that the node will no longer be able to participate in the verification of model reasoning results as a verification node.

[0159] The embodiments of the present application ensure the normal operation of the blockchain system and the security of data by identifying and processing malicious nodes. By canceling the computing power or verification capability of malicious nodes, the system can adjust malicious nodes and improve the overall security of the system. By canceling the capabilities of malicious nodes, the system can regulate user behavior, avoid malicious users or unreliable users abusing computing resources, and ensure the fairness and security of the system. By promptly processing malicious nodes, the stability and reliability of the system are ensured, and the inability to execute tasks due to node failures or malicious behavior is avoided.

[0160] The embodiment of the present application replaces mathematical problems with model reasoning tasks. By assigning model reasoning tasks to computing nodes, computing nodes can perform valuable computing tasks, improve the utilization efficiency of computing resources, and avoid waste of resources. In addition, the selected target computing node is responsible for generating the target block, which ensures the fairness and transparency of the block generation process, further enhances the credibility of blockchain data, and stores the model reasoning results in the target block of the blockchain. Due to the characteristics of the blockchain, these data cannot be tampered with once recorded, ensuring the credibility and security of the model reasoning results. In addition, the target computing node is determined according to the number of model reasoning results corresponding to the computing node, avoiding the situation where a single computing node monopolizes computing resources.

[0161] It should be noted that, for the method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present application are not limited by the described order of actions, because according to the embodiments of the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present application.

[0162] On the basis of the above embodiments, this embodiment further provides a device for executing a model reasoning task, which is applied to electronic devices such as terminal devices and servers.

[0163] Reference Figure 6 , shows a structural block diagram of an embodiment of a model reasoning task execution device of the present application, which is applied to a blockchain system. The blockchain system includes a blockchain and a user, and may specifically include the following modules:

[0164] The model reasoning task acquisition module 601 is used to acquire the model reasoning task initiated by the user; the model reasoning task has corresponding task data;

[0165] A computing node determination module 602 is used to determine a computing node for executing the model reasoning task;

[0166] The model reasoning task execution module 603 is used to send the task data corresponding to the model reasoning task to the computing node, and control the computing node to execute the model reasoning task according to the task data to obtain the model reasoning result corresponding to the computing node;

[0167] A target quantity statistics module 604 is used to count the number of model inference results corresponding to the computing nodes to obtain the target quantity corresponding to the computing nodes;

[0168] A target computing node determination module 605 is used to determine the target computing node according to the target number corresponding to the computing node;

[0169] The target block generation module 606 is used to control the target computing node to generate a target block in the blockchain, and store the model inference result corresponding to the target computing node in the target block.

[0170] The embodiment of the present application also provides a non-volatile readable storage medium, which stores one or more modules (programs). When the one or more modules are applied to a device, the device can execute instructions (instructions) of each method step in the embodiment of the present application.

[0171] The present application embodiment provides one or more machine-readable media on which instructions are stored, and when executed by one or more processors, an electronic device executes one or more of the methods described in the above embodiments. In the present application embodiment, the electronic device includes various types of devices such as terminal devices and servers (clusters).

[0172] The embodiments of the present disclosure may be implemented as a device configured as desired using any appropriate hardware, firmware, software, or any combination thereof, and the device may include electronic devices such as terminal devices and servers (clusters). Figure 7 An exemplary apparatus 700 that can be used to implement various embodiments described in this application is schematically illustrated.

[0173] For one embodiment, Figure 7 An exemplary apparatus 700 is shown having one or more processors 702, a control module (chip set) 704 coupled to at least one of the processor(s) 702, a memory 706 coupled to the control module 704, a non-volatile memory (NVM) / storage device 708 coupled to the control module 704, one or more input / output devices 710 coupled to the control module 704, and a network interface 712 coupled to the control module 704.

[0174] The processor 702 may include one or more single-core or multi-core processors, and the processor 702 may include any combination of general-purpose processors or special-purpose processors (such as graphics processors, application processors, baseband processors, etc.). In some embodiments, the device 700 can be used as a terminal device, server (cluster), etc. described in the embodiments of the present application.

[0175] In some embodiments, the apparatus 700 may include one or more computer-readable media (e.g., memory 706 or NVM / storage device 708) having instructions 714 and one or more processors 702 combined with the one or more computer-readable media and configured to execute the instructions 714 to implement a module to perform the actions described in the present disclosure.

[0176] For one embodiment, the control module 704 may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) 702 and / or any suitable device or component in communication with the control module 704 .

[0177] The control module 704 may include a memory controller module to provide an interface to the memory 706. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0178] The memory 706 may be used, for example, to load and store data and / or instructions 714 for the device 700. For one embodiment, the memory 706 may include any suitable volatile memory, such as a suitable DRAM. In some embodiments, the memory 706 may include double data rate type four synchronous dynamic random access memory (DDR4 SDRAM).

[0179] For one embodiment, control module 704 may include one or more input / output controllers to provide an interface to NVM / storage device 708 and input / output device(s) 710 .

[0180] For example, NVM / storage 708 may be used to store data and / or instructions 714. NVM / storage 708 may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable non-volatile storage device(s) (e.g., one or more hard disk drives (HDDs), one or more compact disk (CD) drives, and / or one or more digital versatile disk (DVD) drives).

[0181] NVM / storage device 708 may include storage resources that are physically part of the device on which apparatus 700 is installed, or it may be accessible to the device without being part of the device. For example, NVM / storage device 708 may be accessed via input / output device(s) 710 over a network.

[0182] (One or more) input / output devices 710 may provide an interface for the apparatus 700 to communicate with any other appropriate device, and the input / output device 710 may include a communication component, an audio component, a sensor component, etc. The network interface 712 may provide an interface for the apparatus 700 to communicate through one or more networks, and the apparatus 700 may wirelessly communicate with one or more components of a wireless network according to any of one or more wireless network standards and / or protocols, for example, accessing a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G, 5G, etc., or a combination thereof for wireless communication.

[0183] For one embodiment, at least one of the processor(s) 702 may be packaged together with the logic of one or more controllers (e.g., a memory controller module) of the control module 704. For one embodiment, at least one of the processor(s) 702 may be packaged together with the logic of one or more controllers of the control module 704 to form a system-in-package (SiP). For one embodiment, at least one of the processor(s) 702 may be integrated on the same die with the logic of one or more controllers of the control module 704. For one embodiment, at least one of the processor(s) 702 may be integrated on the same die with the logic of one or more controllers of the control module 704 to form a system-on-chip (SoC).

[0184] In various embodiments, the apparatus 700 may be, but is not limited to, a terminal device such as a server, a desktop computing device, or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). In various embodiments, the apparatus 700 may have more or fewer components and / or a different architecture. For example, in some embodiments, the apparatus 700 includes one or more cameras, a keyboard, a liquid crystal display (LCD) screen (including a touch screen display), a non-volatile memory port, multiple antennas, a graphics chip, an application-specific integrated circuit (ASIC), and a speaker.

[0185] Among them, the main control chip can be used as a processor or control module in the detection device, sensor data, location information, etc. are stored in a memory or NVM / storage device, the sensor group can be used as an input / output device, and the communication interface may include a network interface.

[0186] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0187] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0188] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable model reasoning task execution terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable model reasoning task execution terminal device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0189] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable model reasoning task execution terminal device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0190] These computer program instructions can also be loaded onto a computer or other programmable terminal device for executing model reasoning tasks, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions executed on the computer or other programmable terminal device for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0191] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0192] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0193] The above is a detailed introduction to a method and device for executing a model reasoning task, an electronic device and a storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for executing a model reasoning task, characterized in that: Applied to a blockchain system, the blockchain system includes a blockchain and a user, and the method includes: Acquire a model reasoning task initiated by the user; the model reasoning task has corresponding task data; Determining a computing node for executing the model reasoning task; Sending the task data corresponding to the model reasoning task to the computing node, and controlling the computing node to perform the model reasoning task according to the task data, to obtain the model reasoning result corresponding to the computing node; Counting the number of model inference results corresponding to the computing nodes to obtain the target number corresponding to the computing nodes; Determine the target computing node according to the target number corresponding to the computing node; The target computing node is controlled to generate a target block in the blockchain, and the model inference result corresponding to the target computing node is stored in the target block.

2. The method according to claim 1, characterized in that The blockchain system also includes an arbitration node. After sending the task data corresponding to the model reasoning task to the computing node and controlling the computing node to perform the model reasoning task according to the task data, and obtaining the model reasoning result corresponding to the computing node, the method further includes: Controlling the computing node to send the model reasoning result to the arbitration node; Determining a verification node for verifying the model reasoning result; Sending the task data corresponding to the model reasoning task to the verification node, and controlling the verification node to verify the model reasoning result according to the task data to obtain a model verification result; Controlling the verification node to send the model verification result to the arbitration node; The arbitration node is controlled to compare whether the model inference result and the model verification result are the same, and obtain an arbitration result for determining whether the computing node is a malicious node; the malicious node is a computing node or a verification node that affects the normal operation of the blockchain system.

3. The method according to claim 2, characterized in that The blockchain system is in communication with a model reasoning task provider, the model reasoning task provider includes a standard reasoning result corresponding to the model reasoning task, and the arbitration node is controlled to compare whether the model reasoning result and the model verification result are the same, and obtain an arbitration result for determining whether the computing node is a malicious node, including: If the model reasoning result and the model verification result are the same, a first arbitration result for determining that the computing node is not the malicious node is obtained; If the model reasoning result and the model verification result are different, obtaining the standard reasoning result corresponding to the model reasoning task from the model reasoning task provider, and obtaining a second arbitration result for determining whether the computing node or the verification node is the malicious node based on the model reasoning result, the model verification result and the standard reasoning result; The first arbitration result or the second arbitration result is determined as the arbitration result.

4. The method according to claim 3, characterized in that The obtaining, according to the model reasoning result, the model verification result and the standard reasoning result, a second arbitration result for determining whether the computing node or the verification node is the malicious node comprises: comparing whether the model reasoning result is the same as the standard reasoning result, and comparing whether the model verification result is the same as the standard reasoning result; If the model reasoning result is different from the standard reasoning result, a third arbitration result for determining that the computing node is the malicious node is obtained; If the model reasoning result is the same as the standard reasoning result, a fourth arbitration result for determining that the computing node is not the malicious node is obtained; If the model verification result is different from the standard reasoning result, obtaining a fifth arbitration result for determining that the verification node is the malicious node; If the model verification result is the same as the standard reasoning result, a sixth arbitration result for determining that the verification node is not the malicious node is obtained; The third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result are determined as the second arbitration result.

5. The method according to claim 1, characterized in that The computing node has a corresponding number of unresponsive times, and the task data corresponding to the model reasoning task is sent to the computing node, and the computing node is controlled to perform the model reasoning task according to the task data to obtain the model reasoning result corresponding to the computing node, including: Sending the task data corresponding to the model reasoning task to the computing node; Determining whether the computing node responds to the model reasoning task within a preset time; If the computing node does not respond to the model inference task within a preset time, the number of non-response times corresponding to the computing node is increased by one.

6. The method according to claim 5, characterized in that The computing node has computing capability for executing the model reasoning task. If the computing node fails to respond to the model reasoning task within a preset time, after the number of non-response times corresponding to the computing node is increased by one, the method further includes: If the number of non-response times corresponding to the computing node exceeds a preset number, the computing capacity of the user corresponding to the computing node is cancelled.

7. The method according to claim 4, characterized in that The computing node has computing capability for executing the model reasoning task, and the verification node has verification capability for verifying the model reasoning result. After determining the third arbitration result and the fifth arbitration result, or the third arbitration result and the sixth arbitration result, or the fourth arbitration result and the fifth arbitration result, or the fourth arbitration result and the sixth arbitration result as the second arbitration result, the method further includes: If the computing node is the malicious node, canceling the computing capability of the computing node; If the verification node is the malicious node, the verification capability of the verification node is cancelled.

8. A device for executing a model reasoning task, characterized in that: Applied to a blockchain system, the blockchain system includes a blockchain and a user, and the device includes: A model reasoning task acquisition module, used to acquire the model reasoning task initiated by the user; the model reasoning task has corresponding task data; A computing node determination module, used to determine a computing node for executing the model reasoning task; A model reasoning task execution module, used to send the task data corresponding to the model reasoning task to the computing node, and control the computing node to execute the model reasoning task according to the task data, so as to obtain the model reasoning result corresponding to the computing node; A target quantity statistics module is used to count the number of model inference results corresponding to the computing nodes to obtain the target quantity corresponding to the computing nodes; A target computing node determination module, used to determine the target computing node according to the target number corresponding to the computing node; The target block generation module is used to control the target computing node to generate a target block in the blockchain, and store the model inference result corresponding to the target computing node in the target block.

9. An electronic device, characterized in that: include: processor; and A memory having executable codes stored thereon, which, when executed, enables the processor to execute the method for executing the model reasoning task as described in any one of claims 1-7.

10. One or more machine-readable media having executable codes stored thereon, which, when executed, enable a processor to execute the method for executing a model reasoning task as described in any one of claims 1-7.

Images