Large model fine-tuning method, device, electronic device, storage medium and program product

By initializing and fixing part of the parameter matrix in a large-scale model and only updating the third parameter matrix, the problem of excessive amount of LoRA fine-tuning parameters and slow fine-tuning speed is solved, and more efficient computing resource utilization and speed improvement is achieved.

CN118364275BActive Publication Date: 2024-11-08MOORE THREADS TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410329948.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-21
Publication Date
2024-11-08
Estimated Expiration
2044-03-21

AI Technical Summary

Technical Problem

LoRA fine-tuning is too large in training parameters in large-scale models, resulting in waste of computing resources and slow fine-tuning.

Method used

By initializing three parameter matrices, the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, and the number of rows and columns of the third parameter matrix is ​​equal to the number of rows and columns of the third parameter matrix is ​​equal to the number of rows and columns of the third parameter matrix is ​​equal to the number of rows and columns of the second parameter matrix is ​​equal to the number of rows and columns of the third parameter matrix is ​​equal to the number of rows and columns of the second parameter matrix. The number of columns of a parameter matrix is ​​fixed during fine-tuning, and the original parameter matrix and the first parameter matrix are updated only to form a new total parameter matrix.

Benefits of technology

The training parameters are significantly reduced, the fine-tuning speed of the large model is improved, and the problem of excessive LoRA fine-tuning parameters is solved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118364275B_ABST
    Figure CN118364275B_ABST
Patent Text Reader

Abstract

The present disclosure relates to the field of electronic digital data processing technology, and in particular to a large model fine-tuning method, device, electronic device, storage medium and program product. The method comprises: initializing a first parameter matrix, a second parameter matrix and a third parameter matrix according to the original parameter matrix corresponding to the target data processing model; fine-tuning the target data processing model using a training sample set to obtain an updated third parameter matrix; determining a newly added parameter matrix corresponding to the target data processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix; determining an updated total parameter matrix according to the original parameter matrix and the newly added parameter matrix. The present disclosure only fine-tunes the third parameter matrix during the fine-tuning of the target data processing model, thereby significantly reducing the amount of training parameters, solving the problem of excessively large parameters in LoRA fine-tuning in certain scenarios, and improving the fine-tuning speed of large models.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the technical field of electrical digital data processing, and in particular to a large model fine-tuning method, a data processing method, a large model fine-tuning device, a data processing device, an electronic device, a storage medium and a program product. Background Art

[0002] Compared with full fine-tuning, the number of training parameters for LoRA fine-tuning is greatly reduced. However, as the scale of random models grows, in many cases, in order to ensure the fine-tuning effect, the rank of LoRA fine-tuning must be set to a larger value. As a result, even with LoRA fine-tuning, the number of training parameters is still large. Summary of the invention

[0003] The present disclosure provides a large model fine-tuning technical solution.

[0004] According to one aspect of the present disclosure, a large model fine-tuning method is provided, comprising:

[0005] Initializing a first parameter matrix, a second parameter matrix, and a third parameter matrix according to the number of rows and columns of the original parameter matrix corresponding to the target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix;

[0006] Fine-tune the target data processing model using a training sample set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target data processing model, the original parameter matrix, the first parameter matrix, and the second parameter matrix remain fixed;

[0007] In response to completion of fine-tuning the target data processing model, determining a newly added parameter matrix corresponding to the target data processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix;

[0008] According to the original parameter matrix and the newly added parameter matrix, an updated total parameter matrix corresponding to the target data processing model is determined.

[0009] In a possible implementation manner, the number of columns of the first parameter matrix is ​​at least one order of magnitude smaller than the number of columns of the original parameter matrix.

[0010] In a possible implementation, the training sample set includes multiple training sample subsets;

[0011] The target data processing model is fine-tuned using the training sample set to obtain an updated third parameter matrix, including:

[0012] Using the multiple training sample subsets to fine-tune the target data processing model respectively, to obtain multiple fine-tuning matrices corresponding to the third parameter matrix;

[0013] An updated third parameter matrix is ​​determined according to the multiple fine-tuning matrices.

[0014] In a possible implementation, the adopting the training sample set to fine-tune the target data processing model to obtain an updated third parameter matrix includes:

[0015] Inputting the training samples in the training sample set into the target data processing model, and outputting the prediction results corresponding to the training samples through the target data processing model;

[0016] Determining a value of a loss function corresponding to the target data processing model according to a prediction result corresponding to the training sample and a label corresponding to the training sample;

[0017] The third parameter matrix is ​​updated according to the value of the loss function.

[0018] In a possible implementation, the step of inputting the training samples in the training sample set into the target data processing model, and outputting the prediction results corresponding to the training samples through the target data processing model includes:

[0019] Determine the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix;

[0020] Determine the sum of the original parameter matrix and the product as the latest total parameter matrix corresponding to the target data processing model;

[0021] The training samples are input into the target data processing model, and based on the latest total parameter matrix, prediction results corresponding to the training samples are obtained.

[0022] In a possible implementation, the target data processing model is a text processing model, and the training sample is a training text.

[0023] According to one aspect of the present disclosure, there is provided a data processing method, comprising:

[0024] Inputting the data to be processed into a target data processing model, and outputting a data processing result corresponding to the data to be processed through the target data processing model;

[0025] Wherein, the target data processing model is obtained based on the large model fine-tuning method.

[0026] In a possible implementation, the data to be processed is text to be processed.

[0027] According to one aspect of the present disclosure, a large model fine-tuning device is provided, comprising:

[0028] An initialization module, used to initialize a first parameter matrix, a second parameter matrix, and a third parameter matrix according to the number of rows and columns of the original parameter matrix corresponding to the target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix;

[0029] A fine-tuning module, configured to fine-tune the target data processing model using a training sample set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target data processing model, the original parameter matrix, the first parameter matrix, and the second parameter matrix remain fixed;

[0030] A first determination module is used to determine, in response to the completion of fine-tuning the target data processing model, a newly added parameter matrix corresponding to the target data processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix;

[0031] The second determination module is used to determine the updated total parameter matrix corresponding to the target data processing model according to the original parameter matrix and the newly added parameter matrix.

[0032] In a possible implementation manner, the number of columns of the first parameter matrix is ​​at least one order of magnitude smaller than the number of columns of the original parameter matrix.

[0033] In a possible implementation, the training sample set includes multiple training sample subsets;

[0034] The fine-tuning module is used to:

[0035] Using the multiple training sample subsets to fine-tune the target data processing model respectively, to obtain multiple fine-tuning matrices corresponding to the third parameter matrix;

[0036] An updated third parameter matrix is ​​determined according to the multiple fine-tuning matrices.

[0037] In a possible implementation, the fine-tuning module is used to:

[0038] Inputting the training samples in the training sample set into the target data processing model, and outputting the prediction results corresponding to the training samples through the target data processing model;

[0039] Determining a value of a loss function corresponding to the target data processing model according to a prediction result corresponding to the training sample and a label corresponding to the training sample;

[0040] The third parameter matrix is ​​updated according to the value of the loss function.

[0041] In a possible implementation, the fine-tuning module is used to:

[0042] Determine the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix;

[0043] Determine the sum of the original parameter matrix and the product as the latest total parameter matrix corresponding to the target data processing model;

[0044] The training samples are input into the target data processing model, and based on the latest total parameter matrix, prediction results corresponding to the training samples are obtained.

[0045] In a possible implementation, the target data processing model is a text processing model, and the training sample is a training text.

[0046] According to one aspect of the present disclosure, there is provided a data processing device, comprising:

[0047] A data processing module, used for inputting the data to be processed into a target data processing model, and outputting a data processing result corresponding to the data to be processed through the target data processing model;

[0048] Wherein, the target data processing model is obtained based on the large model fine-tuning device.

[0049] In a possible implementation, the data to be processed is text to be processed.

[0050] According to one aspect of the present disclosure, an electronic device is provided, comprising: one or more processors; a memory for storing executable instructions; wherein the one or more processors are configured to call the executable instructions stored in the memory to execute the above method.

[0051] According to one aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored, and the computer program instructions implement the above method when executed by a processor.

[0052] According to one aspect of the present disclosure, a computer program product is provided, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in an electronic device, a processor in the electronic device executes the above method.

[0053] In an embodiment of the present disclosure, a first parameter matrix, a second parameter matrix, and a third parameter matrix are initialized according to the number of rows and columns of an original parameter matrix corresponding to a target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix, and the target data processing model is fine-tuned using a training sample set to obtain an updated third parameter matrix, wherein, in the process of fine-tuning the target data processing model, the original parameter matrix, the first The parameter matrix and the second parameter matrix remain fixed, and in response to the completion of fine-tuning of the target data processing model, the newly added parameter matrix corresponding to the target data processing model is determined according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix, and the updated total parameter matrix corresponding to the target data processing model is determined according to the original parameter matrix and the newly added parameter matrix. Thus, on the basis of the first parameter matrix and the second parameter matrix of LoRA fine-tuning, a third parameter matrix with a smaller parameter amount is introduced, and in the process of fine-tuning the target data processing model, only the third parameter matrix is ​​fine-tuned, thereby significantly reducing the amount of training parameters, solving the problem of excessive amount of parameters in LoRA fine-tuning in certain scenarios, and improving the fine-tuning speed of large models.

[0054] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

[0055] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used to illustrate the technical solutions of the present disclosure together with the specification.

[0057] Figure 1 A flowchart of a large model fine-tuning method provided by an embodiment of the present disclosure is shown.

[0058] Figure 2A schematic diagram showing the value of the loss function in the process of fine-tuning the Baichuan 2-13B model using the large model fine-tuning method provided in an embodiment of the present disclosure.

[0059] Figure 3 A block diagram of a large model fine-tuning device provided by an embodiment of the present disclosure is shown.

[0060] Figure 4 A block diagram of an electronic device 1900 provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0061] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0062] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0063] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0064] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0065] The disclosed embodiment provides a large model fine-tuning method, by initializing a first parameter matrix, a second parameter matrix, and a third parameter matrix according to the number of rows and columns of an original parameter matrix corresponding to a target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix, and the target data processing model is fine-tuned using a training sample set to obtain an updated third parameter matrix, wherein, during the fine-tuning of the target data processing model, the original parameter matrix , the first parameter matrix and the second parameter matrix remain fixed, in response to the completion of fine-tuning of the target data processing model, the newly added parameter matrix corresponding to the target data processing model is determined according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix, and the updated total parameter matrix corresponding to the target data processing model is determined according to the original parameter matrix and the newly added parameter matrix, thereby introducing a third parameter matrix with a smaller parameter amount on the basis of the first parameter matrix and the second parameter matrix of LoRA fine-tuning, and in the process of fine-tuning the target data processing model, only the third parameter matrix is ​​fine-tuned, thereby significantly reducing the amount of training parameters, solving the problem of excessive amount of parameters of LoRA fine-tuning in some scenarios, and improving the fine-tuning speed of large models.

[0066] The large model fine-tuning method provided by the embodiment of the present disclosure is described in detail below with reference to the accompanying drawings.

[0067] Figure 1 A flow chart of a large model fine-tuning method provided by an embodiment of the present disclosure is shown. In one possible implementation, the execution subject of the large model fine-tuning method may be a large model fine-tuning device. For example, the large model fine-tuning method may be executed by a terminal device or a server or other electronic device. The terminal device may be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device or a wearable device, etc. In some possible implementations, the large model fine-tuning method may be implemented by a processor calling computer-readable instructions stored in a memory. For example Figure 1 As shown, the large model fine-tuning method includes steps S11 to S14.

[0068] In step S11, the first parameter matrix, the second parameter matrix and the third parameter matrix are initialized according to the number of rows and columns of the original parameter matrix corresponding to the target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix.

[0069] In step S12, the target data processing model is fine-tuned using a training sample set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target data processing model, the original parameter matrix, the first parameter matrix and the second parameter matrix remain fixed.

[0070] In step S13, in response to the completion of fine-tuning the target data processing model, a newly added parameter matrix corresponding to the target data processing model is determined according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix.

[0071] In step S14, an updated total parameter matrix corresponding to the target data processing model is determined according to the original parameter matrix and the newly added parameter matrix.

[0072] The embodiments of the present disclosure can be applied to technical fields such as text, voice, image, video, multimodality, etc. Moreover, the target data processing model in the embodiments of the present disclosure can be used for AIGC (Artificial Intelligence Generative Content), etc., which is not limited here.

[0073] In a possible implementation, the target data processing model is a text processing model, and the training sample is a training text. When the disclosed embodiment is applied to the field of text processing technology, the target data processing model may be a target text processing model, and the training sample may be a training text. After the fine-tuning of the target text processing model is completed, the input of the target text processing model may be the text to be processed.

[0074] For example, the target text processing model can adopt a network structure such as Baichuan 2-13B, ChatGLM-6B, LLaMA-13B, BLOOM, etc., which is not limited here. The text processing model can also be called a language model, etc., which is not limited here.

[0075] The target text processing model can be used in application areas such as dialogue systems (such as medical question and answer), text generation, language error correction, text classification, information retrieval and search engines, speech recognition, speech synthesis, automatic summarization, machine translation, intelligent text analysis, etc., without limitation here.

[0076] In this implementation, the first parameter matrix, the second parameter matrix and the third parameter matrix are initialized according to the number of rows and columns of the original parameter matrix corresponding to the target text processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix, and the target text processing model is fine-tuned using a training text set to obtain an updated third parameter matrix, wherein, during the fine-tuning of the target text processing model, the original parameter matrix, the first parameter matrix and the second parameter matrix remain fixed, in response to the completion of fine-tuning of the target text processing model, the newly added parameter matrix corresponding to the target text processing model is determined according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix, and the updated total parameter matrix corresponding to the target text processing model is determined according to the original parameter matrix and the newly added parameter matrix, thereby introducing a third parameter matrix with a smaller parameter amount on the basis of the first parameter matrix and the second parameter matrix of LoRA fine-tuning, and in the process of fine-tuning the target text processing model, only the third parameter matrix is ​​fine-tuned, thereby significantly reducing the amount of training parameters, solving the problem of excessive amount of parameters of LoRA fine-tuning in some scenarios, and improving the fine-tuning speed of the target text processing model.

[0077] When the embodiments of the present disclosure are applied to the field of speech processing technology, the target data processing model may be a target speech processing model, and the training sample may be a training speech. After the target speech processing model is fine-tuned, the input of the target speech processing model may be a speech to be processed.

[0078] When the embodiments of the present disclosure are applied to the field of image processing technology, the target data processing model may be a target image processing model, and the training samples may be training images. After the target image processing model is fine-tuned, the input of the target image processing model may be an image to be processed.

[0079] When the embodiments of the present disclosure are applied to the field of video processing technology, the target data processing model may be a target video processing model, and the training sample may be a training video. After the target video processing model is fine-tuned, the input of the target video processing model may be a video to be processed.

[0080] When the embodiments of the present disclosure are applied to the field of multimodal processing technology, the target data processing model can be a target multimodal processing model, and the training samples can be training multimodal data. After the target multimodal processing model is fine-tuned, the input of the target multimodal processing model can be the multimodal data to be processed.

[0081] In the embodiment of the present disclosure, the target data processing model may be a large model. For example, the target data processing model may be an LLM (Large Language Model).

[0082] In the embodiments of the present disclosure, the domain corresponding to the source data processing model may be referred to as the source domain, and the domain corresponding to the target data processing model may be referred to as the target domain. The network structure of the target data processing model may be the same as the network structure of the source data processing model, and the target data processing model may be fine-tuned based on the source data processing model.

[0083] In the embodiment of the present disclosure, the original parameter matrix may represent the matrix corresponding to all parameters in the target data processing model (i.e., the source data processing model) before fine-tuning. That is, in the embodiment of the present disclosure, the original parameter matrix may include all parameters in the target data processing model before fine-tuning. In one example, the original parameter matrix may be represented by Wold.

[0084] In the disclosed embodiment, the first parameter matrix, the second parameter matrix, and the third parameter matrix can be initialized according to the number of rows and columns of the original parameter matrix, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix. In one example, the first parameter matrix can be represented by A, the second parameter matrix can be represented by B, and the third parameter matrix can be represented by D.

[0085] For example, the size of the original parameter matrix is ​​m×n, that is, the number of rows of the original parameter matrix is ​​m and the number of columns is n. Then, the size of the first parameter matrix can be m×r, that is, the number of rows of the first parameter matrix can be m and the number of columns can be r. The size of the second parameter matrix can be r×n, that is, the number of rows of the second parameter matrix can be r and the number of columns can be n. The size of the third parameter matrix can be r×r, that is, the number of rows and the number of columns of the third parameter matrix can both be r. Among them, r<m, and r<n.

[0086] In a possible implementation manner, the number of columns of the first parameter matrix is ​​at least one order of magnitude smaller than the number of columns of the original parameter matrix.

[0087] For example, the number of columns of the first parameter matrix can be one order of magnitude smaller than that of the original parameter matrix; for another example, the number of columns of the first parameter matrix can be two orders of magnitude smaller than that of the original parameter matrix; and so on. For example, if the size of the original parameter matrix is m×n, the size of the first parameter matrix is m×r, the size of the second parameter matrix is r×n, and the size of the third parameter matrix is r×r, where r << m and r << n. For example, m = 15360, n = 5120, and r can be 1024 or 512, etc., which is not limited herein.

[0088] In this implementation manner, by setting the number of columns of the first parameter matrix (equal to the number of rows of the second parameter matrix, the number of rows of the third parameter matrix, and the number of columns of the third parameter matrix) to be at least one order of magnitude smaller than the number of rows and columns of the original parameter matrix, it helps to save a large amount of computing resources.

[0089] In the embodiments of the present disclosure, the first parameter matrix, the second parameter matrix, and the third parameter matrix can be initialized by initialization methods such as random initialization and Gaussian distribution initialization, which is not limited herein.

[0090] In a possible implementation manner, the first parameter matrix, the second parameter matrix, and the third parameter matrix can be initialized by Gaussian distribution to increase the generalization ability of the target data processing model.

[0091] In the embodiments of the present disclosure, data in the target domain can be collected, and the data in the target domain can be preprocessed to obtain a training sample set in the target domain, where the training sample set can include multiple training samples. After obtaining the training sample set in the target domain, the target data processing model can be fine-tuned using the training sample set in the target domain.

[0092] In the embodiments of the present disclosure, during the process of fine-tuning the target data processing model, the original parameter matrix, the first parameter matrix, and the second parameter matrix can be fixed, and only the third parameter matrix is fine-tuned. That is, during the process of fine-tuning the target data processing model, the original parameter matrix, the first parameter matrix, and the second parameter matrix can be kept frozen and not further adjusted. By keeping the original parameter matrix, the first parameter matrix, and the second parameter matrix fixed during the fine-tuning process of the target data processing model, a large amount of video memory resources can be saved.

[0093] In a possible implementation, the training samples in the training sample set are input into the target data processing model, and the prediction results corresponding to the training samples are output through the target data processing model, including: determining the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix; determining the sum of the original parameter matrix and the product as the latest total parameter matrix corresponding to the target data processing model; inputting the training samples into the target data processing model, and obtaining the prediction results corresponding to the training samples based on the latest total parameter matrix.

[0094] In this implementation, during the fine-tuning of the target data processing model, Wold+A×D×B can be used as the latest total parameter matrix to participate in the calculation of forward propagation and backward propagation. Wherein, Wold can represent the original parameter matrix, A can represent the first parameter matrix, D can represent the latest third parameter matrix, and B can represent the second parameter matrix. A×D×B can represent the product. For example, the prediction result output by the target data processing model can be Wold×X+A×D×B×X, where X can represent the input of the target data processing model (such as a training sample).

[0095] In this implementation, by determining the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix, the sum of the original parameter matrix and the product is determined as the latest total parameter matrix corresponding to the target data processing model, and the training samples are input into the target data processing model, and based on the latest total parameter matrix, the prediction results corresponding to the training samples are obtained. Therefore, in the process of fine-tuning the target data processing model, the forward propagation calculation can be realized based on the latest total parameter matrix obtained by the latest third parameter matrix.

[0096] In one possible implementation, the target data processing model is fine-tuned using a training sample set to obtain an updated third parameter matrix, including: inputting training samples in the training sample set into the target data processing model, and outputting prediction results corresponding to the training samples through the target data processing model; determining the value of the loss function corresponding to the target data processing model based on the prediction results corresponding to the training samples and the labels corresponding to the training samples; and updating the third parameter matrix based on the value of the loss function.

[0097] In this implementation, for any training sample, the training sample can be input into the target data processing model, the feature vector corresponding to the training sample can be extracted through the target data processing model, and the prediction result corresponding to the training sample can be obtained based on the feature vector corresponding to the training sample.

[0098] A preset loss function (eg, a cross entropy loss function) may be used to determine the value of the loss function according to the prediction result corresponding to the training sample and the label corresponding to the training sample.

[0099] In this implementation, the gradient of the latest total parameter matrix can be determined according to the value of the loss function, the gradient of the third parameter matrix can be determined according to the gradient of the latest total parameter matrix, and the third parameter matrix can be updated according to the gradient of the third parameter matrix. Thus, in the process of fine-tuning the target data processing model, the calculation of back propagation is realized based on the latest total parameter matrix.

[0100] In a possible implementation, fine-tuning may be stopped in response to the value of the loss function corresponding to the target data processing model being less than a preset value. The preset value may be set to a smaller value. For example, the preset value may be set to 0.7, etc., which is not limited here. In this implementation, in the process of fine-tuning the target data processing model, overfitting may be allowed so as to fully fine-tune the target data processing model using the training sample set.

[0101] In one possible implementation, the training sample set includes multiple training sample subsets; and the method of fine-tuning the target data processing model using the training sample set to obtain an updated third parameter matrix includes: fine-tuning the target data processing model using the multiple training sample subsets respectively to obtain multiple fine-tuning matrices corresponding to the third parameter matrix; and determining the updated third parameter matrix based on the multiple fine-tuning matrices.

[0102] As an example of this implementation, an average value of the multiple fine-tuning matrices may be calculated to obtain an updated third parameter matrix.

[0103] As an example of this implementation, a weighted sum of the multiple fine-tuning matrices may be calculated to obtain an updated third parameter matrix, wherein the weight of any fine-tuning matrix among the multiple fine-tuning matrices may be positively correlated with the number of training samples in the training sample subset corresponding to the fine-tuning matrix.

[0104] In this implementation, the training sample set is divided into multiple training sample subsets, and the multiple training sample subsets are used to fine-tune the target data processing model respectively, to obtain multiple fine-tuning matrices corresponding to the third parameter matrix, and based on the multiple fine-tuning matrices, an updated third parameter matrix is ​​determined, thereby helping to improve the fine-tuning effect of the target data processing model.

[0105] In another possible implementation, the target data processing model may be fine-tuned multiple times using the same training sample set to obtain multiple fine-tuning matrices corresponding to the third parameter matrix, and an updated third parameter matrix may be determined based on the multiple fine-tuning matrices.

[0106] In the disclosed embodiment, after the target data processing model stops fine-tuning, the first parameter matrix A, the updated third parameter matrix D and the second parameter matrix B can be multiplied to obtain a newly added parameter matrix Wnew=A×D×B corresponding to the target data processing model. The sum of the original parameter matrix and the newly added parameter matrix can be determined as the updated total parameter matrix corresponding to the target data processing model. For example, the updated total parameter matrix is ​​Wold+Wnew.

[0107] The following is an explanation of the large model fine-tuning method provided by the embodiment of the present disclosure through a specific application scenario. In this application scenario, the target data processing model is a text processing model, the training sample is a training text, and the target text processing model can adopt the network structure of Baichuan 2-13B. The optimizer can use Adam, and the loss function can use the default loss function of the model (such as the cross entropy loss function). Other parameters may include: learning rate = 0.0001, seq_len = 1024, batchsize (batch size) = 16, epoch (cycle) = 1000, rank (rank) = 1024. Among them, seq_len represents the length of the input sequence of the target text processing model.

[0108] In this application scenario, the target domain (target task) can be to generate story types based on children's stories. For example, a training text in the training text set of the target domain is:

[0109]

[0110]

[0111] Among them, "fable" is the label of the training text.

[0112] In this application scenario, the size of the original parameter matrix Wold of the target text processing model can be 15360×5120, the size of the first parameter matrix A can be 15360×1024, the size of the second parameter matrix B can be 1024×5120, and the size of the third parameter matrix D can be 1024×1024. The first parameter matrix A and the second parameter matrix B can be initialized using the Gaussian initialization method.

[0113] Figure 2 A schematic diagram showing the value of the loss function in the process of fine-tuning the Baichuan 2-13B model using the large model fine-tuning method provided by the embodiment of the present disclosure. Figure 2 In , the horizontal axis represents epoch, and the vertical axis represents the value of the loss function. Figure 2 As shown in the figure, when training to the 500th epoch, the value of the loss function roughly converges to 0.6, and the training can be stopped at this time.

[0114] In this application scenario, the training text set of the target domain can be divided into multiple training text sets, and the target text processing model can be fine-tuned using the multiple training text sets to obtain multiple fine-tuning matrices corresponding to the third parameter matrix. The average value of the multiple fine-tuning matrices can be calculated to obtain an updated third parameter matrix.

[0115] The first parameter matrix A, the updated third parameter matrix D and the second parameter matrix B can be multiplied to obtain a newly added parameter matrix Wnew=A×D×B corresponding to the target text processing model. The sum of the original parameter matrix and the newly added parameter matrix can be determined as the updated total parameter matrix corresponding to the target text processing model. For example, the updated total parameter matrix is ​​Wold+Wnew.

[0116] The Baichuan 2-13B model has a total of 40 layers. If LoRA fine-tuning is used, the number of parameters required for fine-tuning is (15360×1024+1024×5120)×40≈839 million. Using the method provided in the embodiment of the present disclosure, the number of parameters required for fine-tuning is 1024×1024×40≈42 million. The embodiment of the present disclosure can significantly reduce the amount of training parameters without sacrificing many effects.

[0117] The disclosed embodiment also provides a data processing method, including: obtaining a target data processing model trained by the large model fine-tuning method; inputting the data to be processed into the target data processing model, and outputting a data processing result corresponding to the data to be processed through the target data processing model.

[0118] In a possible implementation, the data to be processed is text to be processed.

[0119] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, the present disclosure will not repeat them. It can be understood by those skilled in the art that in the above-mentioned method of the specific implementation method, the specific execution order of each step should be determined according to its function and possible internal logic.

[0120] In addition, the present disclosure also provides a large model fine-tuning device, a data processing device, an electronic device, a computer-readable storage medium, and a computer program product, all of which can be used to implement any large model fine-tuning method or data processing method provided by the present disclosure. The corresponding technical solutions and technical effects can be found in the corresponding records in the method part and will not be repeated here.

[0121] Figure 3 FIG. 1 is a block diagram of a large model fine-tuning device provided by an embodiment of the present disclosure. Figure 3 As shown, the large model fine-tuning device includes:

[0122] An initialization module 31 is used to initialize a first parameter matrix, a second parameter matrix and a third parameter matrix according to the number of rows and columns of the original parameter matrix corresponding to the target data processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix;

[0123] A fine-tuning module 32, configured to fine-tune the target data processing model using a training sample set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target data processing model, the original parameter matrix, the first parameter matrix and the second parameter matrix remain fixed;

[0124] A first determination module 33 is used to determine, in response to the completion of fine-tuning of the target data processing model, a newly added parameter matrix corresponding to the target data processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix;

[0125] The second determination module 34 is used to determine the updated total parameter matrix corresponding to the target data processing model according to the original parameter matrix and the newly added parameter matrix.

[0126] In a possible implementation manner, the number of columns of the first parameter matrix is ​​at least one order of magnitude smaller than the number of columns of the original parameter matrix.

[0127] In a possible implementation, the training sample set includes multiple training sample subsets;

[0128] The fine-tuning module 32 is used for:

[0129] Using the multiple training sample subsets to fine-tune the target data processing model respectively, to obtain multiple fine-tuning matrices corresponding to the third parameter matrix;

[0130] An updated third parameter matrix is ​​determined according to the multiple fine-tuning matrices.

[0131] In a possible implementation, the fine-tuning module 32 is used to:

[0132] Inputting the training samples in the training sample set into the target data processing model, and outputting the prediction results corresponding to the training samples through the target data processing model;

[0133] Determining a value of a loss function corresponding to the target data processing model according to a prediction result corresponding to the training sample and a label corresponding to the training sample;

[0134] The third parameter matrix is ​​updated according to the value of the loss function.

[0135] In a possible implementation, the fine-tuning module 32 is used to:

[0136] Determine the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix;

[0137] Determine the sum of the original parameter matrix and the product as the latest total parameter matrix corresponding to the target data processing model;

[0138] The training samples are input into the target data processing model, and based on the latest total parameter matrix, prediction results corresponding to the training samples are obtained.

[0139] In a possible implementation, the target data processing model is a text processing model, and the training sample is a training text.

[0140] The present disclosure also provides a data processing device, including:

[0141] A data processing module, used for inputting the data to be processed into a target data processing model, and outputting a data processing result corresponding to the data to be processed through the target data processing model;

[0142] Wherein, the target data processing model is obtained based on the large model fine-tuning device.

[0143] In a possible implementation, the data to be processed is text to be processed.

[0144] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. Its specific implementation and technical effects can refer to the description of the above method embodiments. For the sake of brevity, they will not be repeated here.

[0145] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions implement the above method when executed by a processor. The computer-readable storage medium may be a non-volatile computer-readable storage medium, or may be a volatile computer-readable storage medium.

[0146] The embodiment of the present disclosure further provides a computer program, including a computer-readable code. When the computer-readable code is executed in an electronic device, a processor in the electronic device executes the above method.

[0147] The embodiments of the present disclosure also provide a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in an electronic device, a processor in the electronic device executes the above method.

[0148] An embodiment of the present disclosure also provides an electronic device, comprising: one or more processors; and a memory for storing executable instructions; wherein the one or more processors are configured to call the executable instructions stored in the memory to execute the above method.

[0149] The electronic device may be provided as a terminal, a server, or a device in other forms.

[0150] Figure 4 1 is a block diagram of an electronic device 1900 provided in an embodiment of the present disclosure. For example, the electronic device 1900 may be provided as a server or a terminal. Figure 4 , the electronic device 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions executable by the processing component 1922, such as an application. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0151] The electronic device 1900 may also include a power supply component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input / output interface 1958 (I / O interface). The electronic device 1900 may operate based on an operating system stored in the memory 1932, such as a Microsoft Server operating system (Windows Server 2003). TM ), a graphical user interface operating system launched by Apple (MacOS X TM ), a multi-user, multi-process computer operating system (Unix TM ), a free and open source Unix-like operating system (Linux TM ), an open source Unix-like operating system (FreeBSD TM ) or similar.

[0152] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions, which can be executed by the processing component 1922 of the electronic device 1900 to perform the above method.

[0153] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0154] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0155] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0156] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0157] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0158] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0159] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0160] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0161] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK) and the like.

[0162] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0163] If the technical solution of the embodiments of the present disclosure involves personal information, the product using the technical solution of the embodiments of the present disclosure has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of the embodiments of the present disclosure involves sensitive personal information, the product using the technical solution of the embodiments of the present disclosure has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the requirement of "explicit consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to collect his or her personal information; or on the device for processing personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

[0164] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A large model fine-tuning method, It is characterized in that include: Initialize the first parameter matrix, the second parameter matrix and the third parameter matrix according to the number of rows and columns of the original parameter matrix corresponding to the target text processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix; Fine-tune the target text processing model using a training text set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target text processing model, the original parameter matrix, the first parameter matrix, and the second parameter matrix remain fixed; In response to completion of fine-tuning the target text processing model, determining a newly added parameter matrix corresponding to the target text processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix; According to the original parameter matrix and the newly added parameter matrix, an updated total parameter matrix corresponding to the target text processing model is determined.

2. The method according to claim 1, It is characterized in that The number of columns of the first parameter matrix is ​​at least one order of magnitude smaller than the number of columns of the original parameter matrix.

3. The method according to claim 1 or 2, It is characterized in that The training text set includes a plurality of training text subsets; The target text processing model is fine-tuned using the training text set to obtain an updated third parameter matrix, including: Using the multiple training text subsets to fine-tune the target text processing model respectively, to obtain multiple fine-tuning matrices corresponding to the third parameter matrix; An updated third parameter matrix is ​​determined according to the multiple fine-tuning matrices.

4. The method according to claim 1 or 2, It is characterized in that The target text processing model is fine-tuned using the training text set to obtain an updated third parameter matrix, including: Inputting the training text in the training text set into the target text processing model, and outputting the prediction result corresponding to the training text through the target text processing model; Determining a value of a loss function corresponding to the target text processing model according to a prediction result corresponding to the training text and a label corresponding to the training text; The third parameter matrix is ​​updated according to the value of the loss function.

5. The method according to claim 4, It is characterized in that The step of inputting the training text in the training text set into the target text processing model and outputting the prediction result corresponding to the training text through the target text processing model includes: Determine the product of the first parameter matrix, the latest third parameter matrix and the second parameter matrix; Determine the sum of the original parameter matrix and the product as the latest total parameter matrix corresponding to the target text processing model; The training text is input into the target text processing model, and based on the latest total parameter matrix, a prediction result corresponding to the training text is obtained.

6. A data processing method, It is characterized in that include: Inputting the text to be processed into a target text processing model, and outputting a data processing result corresponding to the text to be processed through the target text processing model; Wherein, the target text processing model is obtained based on the large model fine-tuning method described in any one of claims 1 to 5.

7. A large model fine-tuning device, It is characterized in that include: An initialization module, used for initializing a first parameter matrix, a second parameter matrix and a third parameter matrix according to the number of rows and columns of the original parameter matrix corresponding to the target text processing model, wherein the number of rows of the first parameter matrix is ​​equal to the number of rows of the original parameter matrix, the number of columns of the second parameter matrix is ​​equal to the number of columns of the original parameter matrix, the number of columns of the first parameter matrix is ​​less than the number of columns of the original parameter matrix, and the number of rows of the second parameter matrix, the number of rows of the third parameter matrix and the number of columns of the third parameter matrix are all equal to the number of columns of the first parameter matrix; A fine-tuning module, used to fine-tune the target text processing model using a training text set to obtain an updated third parameter matrix; wherein, during the fine-tuning of the target text processing model, the original parameter matrix, the first parameter matrix and the second parameter matrix remain fixed; A first determination module is used to determine, in response to the completion of fine-tuning of the target text processing model, a newly added parameter matrix corresponding to the target text processing model according to the first parameter matrix, the second parameter matrix and the updated third parameter matrix; The second determination module is used to determine the updated total parameter matrix corresponding to the target text processing model according to the original parameter matrix and the newly added parameter matrix.

8. A data processing device, It is characterized in that include: A data processing module, used for inputting the text to be processed into a target text processing model, and outputting a data processing result corresponding to the text to be processed through the target text processing model; Wherein, the target text processing model is obtained based on the large model fine-tuning device described in claim 7.

9. An electronic device, It is characterized in that include: one or more processors; a memory for storing executable instructions; The one or more processors are configured to call executable instructions stored in the memory to execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having computer program instructions stored thereon, It is characterized in that When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

11. A computer program product comprising computer readable code, or a non-volatile computer readable storage medium carrying computer readable code, It is characterized in that When the computer readable code is executed in an electronic device, a processor in the electronic device executes the method according to any one of claims 1 to 6.