Teaching data evaluation method and device, equipment and storage medium

Abstract

The invention discloses a teaching data evaluation method and device, equipment and a storage medium. The method comprises the steps of obtaining to-be-evaluated teaching data; determining a target sample set according to the to-be-evaluated teaching data; training an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model; and evaluating the target VS-GPT model to obtain an evaluation result corresponding to the teaching data to be evaluated. Through the technical scheme of the invention, the collected teaching data can be used as input, the GPT model is subjected to self-supervised training, the trained VS-GPT model is used as a virtual student, and the teaching data is evaluated by evaluating the virtual student, so that the trial and error cost is reduced.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of artificial intelligence technology, and in particular to a teaching data evaluation method, device, equipment and storage medium. Background Art

[0002] The exploration of educational models is a long-term and far-reaching task. The current educational model relies on the exploration and experience of predecessors, but with the development of the times, it is imperative to explore new educational models. If we still rely on traditional methods and means, such as accumulating and summarizing through school pilots, evaluating each type of teaching data, and then obtaining better teaching data, it will seriously lag behind the needs of the development of the times, and the cost of trial and error is very high. Summary of the invention

[0003] The embodiments of the present invention provide a teaching data evaluation method, device, equipment and storage medium, so as to realize the self-supervised training of the GPT model by using the collected teaching data as input, and the trained VS-GPT model as a virtual student. The teaching data is evaluated by evaluating the virtual student, thereby reducing the trial and error cost.

[0004] According to one aspect of the present invention, a teaching data evaluation method is provided, comprising:

[0005] Obtain the teaching data to be evaluated;

[0006] Determine a target sample set according to the teaching data to be evaluated;

[0007] Training an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model;

[0008] The target VS-GPT model is evaluated to obtain an evaluation result corresponding to the teaching data to be evaluated.

[0009] According to another aspect of the present invention, a teaching data evaluation device is provided, the device comprising:

[0010] An acquisition module is used to obtain the teaching data to be evaluated;

[0011] A determination module, used to determine a target sample set according to the teaching data to be evaluated;

[0012] A training module, used to train an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model;

[0013] The evaluation module is used to evaluate the target VS-GPT model and obtain the evaluation results corresponding to the teaching data to be evaluated.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0015] at least one processor; and

[0016] a memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the teaching data evaluation method described in any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the teaching data evaluation method described in any embodiment of the present invention when executed.

[0019] The embodiment of the present invention first obtains the teaching data to be evaluated, determines the target sample set according to the teaching data to be evaluated, and then uses the target sample set as input to perform self-supervised training on the initial VS-GPT model to obtain a trained target VS-GPT model, and then evaluates the target VS-GPT model to obtain the evaluation results corresponding to the teaching data to be evaluated. Through the technical solution of the present invention, the collected teaching data can be used as input to perform self-supervised training on the initial VS-GPT model, and the trained target VS-GPT model can be used as a virtual student. The teaching data can be evaluated by evaluating the virtual student, thereby reducing the cost of trial and error.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 is a flow chart of a teaching data evaluation method in an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of an initial VS-GPT model in an embodiment of the present invention;

[0024] Figure 3 is a structural diagram of a teaching data evaluation device in an embodiment of the present invention;

[0025] Figure 4 It is a structural schematic diagram of an electronic device for implementing the teaching data evaluation method of an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Embodiment 1

[0029] Figure 1 is a flow chart of a teaching data evaluation method in an embodiment of the present invention. This embodiment is applicable to teaching data evaluation. The method can be executed by a teaching data evaluation device in an embodiment of the present invention. The device can be implemented in software and / or hardware. Figure 1 As shown, the method specifically comprises the following steps:

[0030] S101. Obtain teaching data to be evaluated.

[0031] In this embodiment, the teaching data to be evaluated may be teaching data such as teaching materials, textbooks, recommended extracurricular reading materials, teacher's handouts, and class schedules for various subjects (such as Chinese, mathematics, or English) used in a certain teaching stage (such as primary school, junior high school, high school, or university).

[0032] Specifically, the teaching data to be evaluated are collected. For example, taking primary education as an example, the teaching data to be evaluated may include: teaching materials and textbooks for various subjects used in primary school, recommended extracurricular reading materials, teacher's handouts, and class timetables, etc.

[0033] S102. Determine a target sample set based on the teaching data to be evaluated.

[0034] It should be noted that the target sample set can be a data sample set sampled from the teaching data to be evaluated, which is used to train the initial VS-GPT model.

[0035] Specifically, after collecting the teaching data to be evaluated, the teaching data to be evaluated is sampled to obtain a target sample set for training the initial VS-GPT model.

[0036] S103. Train an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model.

[0037] In this embodiment, the initial VS-GPT model can be an untrained initial VS-GPT model. It can be known that GPT (Generative Pre-trained Transformer) is a pre-trained language model that uses the Transformer architecture to process NLP (Natural Language Processing) tasks. GPT can generate human-readable natural language texts, such as conversations, articles, or news reports. By using large-scale pre-training data and autonomous learning technology, GPT can capture grammatical rules, common expressions, common contexts, and semantic meanings in natural language.

[0038] The target VS-GPT model may be a trained VS-GPT model obtained by training the initial VS-GPT model using a self-supervised method according to the target sample set. In this embodiment, the trained target VS-GPT model may be used as a virtual student.

[0039] Specifically, the initial VS-GPT model is trained using a self-supervised method based on the target sample set to obtain a trained target VS-GPT model, that is, a virtual student is trained.

[0040] S104. Evaluate the target VS-GPT model to obtain evaluation results corresponding to the teaching data to be evaluated.

[0041] In this embodiment, the evaluation result may be the evaluation result of the teaching materials, textbooks, recommended extracurricular readings, teacher's handouts, and class schedules of each subject used in the teaching stage corresponding to the teaching data to be evaluated. Exemplarily, the test result may be, for example, the score corresponding to each type of teaching data to be evaluated.

[0042] Specifically, the target VS-GPT model, i.e., the virtual student, is evaluated to obtain the evaluation results corresponding to the teaching data to be evaluated, such as the scores corresponding to each type of teaching data to be evaluated. After obtaining the evaluation results, such as the scores corresponding to each type of teaching data to be evaluated, the school can select the textbooks, recommended extracurricular readings, teacher handouts, and timetables for each subject used in the teaching stage corresponding to the teaching data to be evaluated based on the evaluation results, i.e., the scores. For example, the textbooks, recommended extracurricular readings, teacher handouts, and timetables for each subject with the highest scores can be used as the optimal textbooks, recommended extracurricular readings, teacher handouts, and timetables for each subject, and then reform and update the teaching model, so as to quickly realize the exploration of the teaching model.

[0043] The embodiment of the present invention first obtains the teaching data to be evaluated, determines the target sample set according to the teaching data to be evaluated, and then uses the target sample set as input to perform self-supervised training on the initial VS-GPT model to obtain a trained target VS-GPT model, and then evaluates the target VS-GPT model to obtain the evaluation results corresponding to the teaching data to be evaluated. Through the technical solution of the present invention, the collected teaching data can be used as input to perform self-supervised training on the initial VS-GPT model, and the trained target VS-GPT model can be used as a virtual student. The teaching data can be evaluated by evaluating the virtual student, thereby reducing the cost of trial and error.

[0044] Optionally, the target sample set is determined based on the teaching data to be evaluated, including:

[0045] Obtain the teaching data to be evaluated.

[0046] The teaching data to be evaluated include: teaching materials to be evaluated and teaching plans to be evaluated.

[0047] In this embodiment, the teaching materials to be evaluated may be teaching materials such as textbooks, recommended extracurricular reading materials, and teacher's handouts for various subjects (such as Chinese, mathematics, or English) used in a certain teaching stage (such as primary school, junior high school, high school, or university); the teaching plan to be evaluated may be a teaching plan such as a class schedule used in a certain teaching stage (such as primary school, junior high school, high school, or university).

[0048] The teaching materials to be evaluated are sampled according to the teaching plan to be evaluated to obtain the target sample set.

[0049] In this embodiment, the teaching materials to be evaluated are sampled according to the teaching plan to be evaluated. For example, the teaching materials of each subject, recommended extracurricular reading materials, and teacher's handouts are sampled according to the class schedule to obtain a target sample set.

[0050] Optionally, the teaching materials to be evaluated are sampled according to the teaching plan to be evaluated to obtain a target sample set, including:

[0051] The proportion information and / or sorting information of each type of teaching materials to be evaluated is determined according to the teaching plan to be evaluated.

[0052] In this embodiment, the teaching materials to be evaluated can be divided into different types. For example, they can be classified according to the types of different subjects. The teaching materials to be evaluated can be divided into various types of teaching materials to be evaluated, such as Chinese teaching materials to be evaluated, mathematics teaching materials to be evaluated, and English teaching materials to be evaluated.

[0053] In this embodiment, the teaching plan to be evaluated may be a class schedule, as shown in Table 1. Generally, a class schedule for a day will be arranged so that different subjects are interspersed. In this case, there will be an arrangement order for each subject and a ratio of the number of classes.

[0054] Table 1

[0055] Monday Tuesday Wednesday Thursday Friday 1 Language math Language English Language 2 math English math Language physical education 3 math Language physical education math English 4 English physical education Language English math 5 Language English English Language Language 6 physical education math Language physical education Language 7 math physical education physical education math physical education

[0056] The ratio information may be the ratio of the number of classes arranged for different subjects in the timetable. For example, the ratio of Chinese, mathematics and English classes arranged in a week in the timetable shown in Table 1 may be 11:9:7.

[0057] The sorting information may be the order information of different subjects in the timetable. For example, in the timetable shown in Table 1, the sorting information of all courses on Monday is Chinese, Mathematics, Mathematics, English, Chinese, Physical Education, Mathematics.

[0058] Specifically, according to the teaching plan to be evaluated, such as a timetable, the proportion information and / or sorting information of each type of teaching materials to be evaluated may be determined.

[0059] Each type of teaching material to be evaluated is sampled according to the proportion information and / or ranking information of each type of teaching material to be evaluated to obtain a target sample set.

[0060] Specifically, each type of teaching material to be evaluated can be sampled according to the ratio information of each type of teaching material to be evaluated to obtain a target sample set. For example, the ratio information of Chinese, mathematics and English can be 11:9:7, then the teaching materials to be evaluated of Chinese, mathematics and English can be sampled according to the ratio of 11:9:7 to obtain a target sample set.

[0061] In addition, each type of teaching material to be evaluated can be sampled according to the sorting information of each type of teaching material to be evaluated to obtain a target sample set. For example, the sorting information of Chinese, mathematics and English can be Chinese, mathematics and English, then the teaching materials to be evaluated of Chinese, mathematics and English can be sampled in the order of Chinese, mathematics and English to obtain a target sample set.

[0062] In addition, each type of teaching materials to be evaluated can be sampled according to the proportion information and / or sorting information of each type of teaching materials to be evaluated to obtain a target sample set. For example, the sorting information of Chinese, mathematics and English classes can be Chinese, mathematics and English, and the specific proportion information of Chinese, mathematics and English classes can be 11:9:7. Then, the teaching materials to be evaluated in the Chinese category, the teaching materials to be evaluated in the mathematics category and the teaching materials to be evaluated in the English category can be sampled in the order of Chinese, mathematics and English and the proportion of 11:9:7 to obtain a target sample set.

[0063] According to each type of teaching material to be evaluated, such as the timetable, each type of teaching material to be evaluated is organized and sampled to generate batch training data, i.e., the target sample set, for training the initial VS-GPT model to simulate the school's teaching process. This is the key to the trained target VS-GPT model to explore the teaching mode. Different organization and sampling methods of training data can obtain different training models, which can simulate the training and learning process of people.

[0064] Optionally, the target VS-GPT model is evaluated to obtain evaluation results corresponding to the teaching data to be evaluated, including:

[0065] Obtain the target question bank corresponding to the teaching data to be evaluated.

[0066] It should be explained that the target question bank can be a question bank used to evaluate the target VS-GPT model, for example, it can be a question bank corresponding to a certain teaching stage (for example, it can be elementary school, junior high school, high school or university).

[0067] Specifically, the target question bank corresponding to the teaching data to be evaluated can be obtained by searching on the Internet, or the target question bank corresponding to the teaching data to be evaluated can be obtained by collecting the target question bank on a pilot basis in schools, or questions can be manually designed as the target question bank.

[0068] The test questions in the target question bank are input into the target VS-GPT model in sequence to obtain the predicted solution information corresponding to each test question.

[0069] The predicted solution information includes: at least one word.

[0070] It should be noted that the target question bank may contain a large number of test questions, and the test questions may be test questions corresponding to various types of teaching materials to be evaluated, such as Chinese test questions corresponding to Chinese teaching materials to be evaluated, mathematics test questions corresponding to mathematics teaching materials to be evaluated, and English test questions corresponding to English teaching materials to be evaluated, etc.

[0071] Among them, the predicted solution information can be the solution information corresponding to each test question obtained by the target VS-GPT model based on the test questions in the target question bank after the test questions in the target question bank are input into the target VS-GPT model in sequence.

[0072] Specifically, the test questions in the target question bank are input into the target VS-GPT model in sequence to obtain the predicted solution information corresponding to each test question given by the target VS-GPT model according to the test questions in the target question bank. In this embodiment, the predicted solution information includes: at least one word.

[0073] The evaluation result corresponding to the teaching data to be evaluated is determined according to the solution information corresponding to each test question and the predicted solution information corresponding to each test question.

[0074] It should be explained that the problem-solving information may be the answer information corresponding to each test question. Specifically, the problem-solving information corresponding to each test question may be the answer information of each test question when obtaining the target question bank corresponding to the teaching data to be evaluated, or the answer information searched on the Internet for each test question, or the answer manually made for each test question.

[0075] Specifically, the solution information corresponding to each test question and the predicted solution information corresponding to each test question can be compared to determine whether the two are the same, or to determine whether the similarity between the two reaches a preset similarity threshold (wherein the preset similarity threshold can be a similarity threshold pre-set by the user according to actual conditions, and this embodiment does not limit this), and then determine the evaluation results corresponding to the teaching data to be evaluated.

[0076] Optionally, obtain a target question bank corresponding to the teaching data to be evaluated, including:

[0077] Obtain an initial question bank corresponding to the teaching material to be evaluated.

[0078] It should be explained that the initial question bank may be a question bank corresponding to a certain teaching stage (for example, primary school, junior high school, high school or university).

[0079] Specifically, the initial question bank corresponding to the teaching materials to be evaluated can be obtained by searching on the Internet, or by collecting and obtaining the initial question bank corresponding to the teaching materials to be evaluated in a school pilot, or by manually designing questions as the initial question bank.

[0080] The initial question bank is screened according to the teaching plan to be evaluated to obtain the target question bank.

[0081] Specifically, the initial question bank can be screened according to the teaching plan to be evaluated, such as the class schedule, to obtain the target question bank. For example, if the class schedule only includes three courses: Chinese, mathematics and English, then the relevant test questions corresponding to Chinese, mathematics and English can be screened from the initial question bank to form the target question bank.

[0082] Optionally, the test questions in the target question bank are input into the target VS-GPT model in sequence to obtain the predicted solution information corresponding to each test question, including:

[0083] The target test question in the target question bank is input into the target VS-GPT model to obtain the first output result corresponding to the target test question.

[0084] The target test question may be a test question in any target question bank.

[0085] It should be noted that the first output result may be the output result corresponding to the target test question obtained by the target VS-GPT model based on the target test question in the target question bank after the target test question in the target question bank is input into the target VS-GPT model.

[0086] Specifically, the target test questions in the target question bank are input into the target VS-GPT model to obtain a first output result corresponding to the target test questions given by the target VS-GPT model based on the target test questions in the target question bank.

[0087] If the first output result corresponding to the target test question does not contain a terminator, the first output result is input into the target VS-GPT model to obtain a second output result corresponding to the target test question.

[0088] In this embodiment, the terminator may be a symbol used to mark the end of the target VS-GPT model evaluation. This embodiment does not limit the form and content of the terminator.

[0089] Among them, the second output result can be the output result corresponding to the target test question obtained by the target VS-GPT model based on the first output result after the first output result is input into the target VS-GPT model.

[0090] Specifically, if the first output result corresponding to the target test question does not contain an end symbol, the evaluation will continue, and the first output result will be input as input into the target VS-GPT model to obtain the second output result corresponding to the target test question given by the target VS-GPT model based on the first output result, until the output result contains an end symbol and the answer is ended.

[0091] If the second output result corresponding to the target test question includes an end symbol, the second output result is determined as the predicted solution information corresponding to the target test question.

[0092] Specifically, if the second output result corresponding to the target test question includes an end symbol, the assessment ends, and the second output result is determined as the predicted solution information corresponding to the target test question.

[0093] Through a series of tests on the target VS-GPT model, we can obtain quantitative evaluation results under the teaching materials and timetable arrangements of the subject. This result can assist schools in the selection of teaching materials and timetables and accelerate the exploration of teaching models.

[0094] Optionally, an initial VS-GPT model is trained according to a target sample set to obtain a target VS-GPT model, including:

[0095] Obtain the word encoding vector, position encoding vector, and preset learnable encoding vector corresponding to the samples in the target sample set.

[0096] The sample may be a sample obtained by sampling each type of teaching material to be evaluated according to the proportion information and / or ranking information of each type of teaching material to be evaluated.

[0097] It should be noted that the word encoding vector may be a word encoding input into the initial VS-GPT model for training the initial VS-GPT model. In this embodiment, the word encoding vector may be represented as word embedding; the preset learnable encoding vector may be a pre-set learnable vector input into the initial VS-GPT model for training the initial VS-GPT model. In this embodiment, the preset learnable encoding vector may be represented as mask embedding. Specifically, the preset learnable encoding vector mask embedding may be used to represent the encoding vector of the masked word; the position encoding vector may be a position encoding input into the initial VS-GPT model for training the initial VS-GPT model. In this embodiment, the position encoding vector may be represented as position embedding. The position encoding vector position embedding may adopt a learnable parameter vector or a fixed position vector of sine-cosine, and its dimension is the same as the word encoding vector word embedding and the preset learnable encoding vector mask embedding.

[0098] Specifically, the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the samples in the target sample set are obtained, wherein the preset learnable encoding vector may also be referred to as the encoding vector of the masked word, that is, when the samples in the target sample set are input into the initial VS-GPT model, it is necessary to perform a mask operation on at least one word in each sample, and the preset learnable encoding vector corresponding to the sample obtained after the mask operation is used to train the initial VS-GPT model.

[0099] The word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample are input into the initial VS-GPT model to obtain the predicted mask text.

[0100] The predicted mask text includes: at least one predicted mask word.

[0101] It should be explained that the predicted mask text can be the mask text predicted by the initial VS-GPT model according to the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample. The predicted mask word can be the mask word included in the mask text predicted by the initial VS-GPT model according to the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample.

[0102] Specifically, when the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample are input into the initial VS-GPT model, each word encoding vector is matched with a corresponding position encoding vector as input, and each preset learnable encoding vector is also matched with a corresponding position encoding vector as input. The initial VS-GPT model predicts a predicted mask text based on the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample, wherein the predicted mask text includes at least one predicted mask word.

[0103] The initial VS-GPT model is trained according to the objective function formed by the predicted mask text and the mask text corresponding to the sample to obtain the target VS-GPT model.

[0104] It should be noted that the objective function can be the loss function of the initial VS-GPT model during self-supervised training.

[0105] In this embodiment, the objective function formed by predicting the mask text and the mask text corresponding to the sample can be expressed as:

[0106]

[0107] Among them, Loss represents the objective function formed by predicting the mask text and the mask text corresponding to the sample, u i represents a predicted mask word in the predicted mask text, m represents the window length corresponding to the predicted mask text, and ui-1 represents a mask word in the mask text corresponding to the sample, k represents the window length corresponding to the sample, and i represents the number of samples in the target sample set.

[0108] Specifically, the initial VS-GPT model is trained according to the objective function formed by the predicted mask text and the mask text corresponding to the sample. Specifically, it can be returning to execute the operation of inputting the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample into the initial VS-GPT model to obtain the predicted mask text until a trained target VS-GPT model is obtained.

[0109] The trained target VS-GPT model is used as a virtual student, and its learning outcomes are used to evaluate the teaching data. In order to unify the self-supervised training and evaluation process, the evaluation can be set in the form of question and answer, and the window length corresponding to the predicted mask text is set to m. Unlike traditional GPT training (m=1), it can be set to m>1 here, which can speed up the training and prediction process.

[0110] Optionally, the initial VS-GPT model includes: at least two tansformer layers and at least one fully connected layer.

[0111] As an exemplary description of an embodiment of the present invention, Figure 2 is a schematic diagram of an initial VS-GPT model in an embodiment of the present invention. Figure 2 As shown, the initial VS-GPT model includes L layers of Tansformer Layer (i.e., tansformer layer) and one Text Prediction layer, where the Text Prediction layer is an fc layer, also known as a fully connected layer, which is used to predict the mask word features output by the last tansformer layer. Each Tansformer Layer (i.e., tansformer layer) includes (from bottom to top): Norm (normalization layer), Multi-Head Attention (multi-attention mechanism module), Norm (normalization layer) and MLP (Multilayer Perceptron).

[0112] Figure 2 In the figure, the black solid long box word embedding represents the word encoding vector, the white hollow long box maskembedding represents the preset learnable encoding vector, and the black and white long box position embedding represents the position encoding vector. Figure 2It can be seen that when the word encoding vector, position encoding vector and preset learnable encoding vector corresponding to the sample are input into the initial VS-GPT model, each word encoding vector is matched with a corresponding position encoding vector as input, and each preset learnable encoding vector is also matched with a corresponding position encoding vector as input. The final target VS-GPT model outputs the predicted mask text.

[0113] The technical solution of the embodiment of the present invention organizes and samples each type of teaching material to be evaluated, such as a timetable, to generate batch training data, i.e., a target sample set, for training an initial VS-GPT model, simulates the school's teaching process, obtains a target VS-GPT model, and uses the target VS-GPT model as a virtual student. By evaluating the target VS-GPT model, i.e., the virtual student, an evaluation result corresponding to the teaching data to be evaluated is obtained. After obtaining the evaluation result, the school can select the textbooks, recommended extracurricular readings, teacher's handouts, and timetables of each subject used in the teaching stage corresponding to the teaching data to be evaluated based on the evaluation result, thereby quickly realizing the exploration of teaching models and reducing the cost of trial and error.

[0114] Embodiment 2

[0115] Figure 3 is a structural diagram of a teaching data evaluation device in an embodiment of the present invention. This embodiment is applicable to teaching data evaluation. The device can be implemented in software and / or hardware. The device can be integrated in any device that provides the function of teaching data evaluation, such as Figure 3 As shown, the teaching data evaluation device specifically includes: an acquisition module 201, a determination module 202, a training module 203 and an evaluation module 204.

[0116] Wherein, the acquisition module 201 is used to acquire the teaching data to be evaluated;

[0117] A determination module 202 is used to determine a target sample set according to the teaching data to be evaluated;

[0118] A training module 203 is used to train an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model;

[0119] The evaluation module 204 is used to evaluate the target VS-GPT model and obtain the evaluation results corresponding to the teaching data to be evaluated.

[0120] Optionally, the determining module 202 includes:

[0121] A first acquisition unit is used to acquire teaching data to be evaluated, wherein the teaching data to be evaluated includes: teaching materials to be evaluated and teaching plans to be evaluated;

[0122] The sampling unit is used to sample the teaching materials to be evaluated according to the teaching plan to be evaluated to obtain a target sample set.

[0123] Optionally, the sampling unit is specifically used for:

[0124] Determine the proportion information and / or sorting information of each type of teaching materials to be evaluated according to the teaching plan to be evaluated;

[0125] Each type of teaching material to be evaluated is sampled according to the proportion information and / or ranking information of each type of teaching material to be evaluated to obtain a target sample set.

[0126] Optionally, the evaluation module 204 includes:

[0127] A second acquisition unit is used to acquire a target question bank corresponding to the teaching data to be evaluated;

[0128] A first input unit is used to input the test questions in the target question bank into the target VS-GPT model in sequence to obtain predicted solution information corresponding to each test question, wherein the predicted solution information includes: at least one word;

[0129] A determination unit is used to determine the evaluation result corresponding to the teaching data to be evaluated according to the solution information corresponding to each test question and the predicted solution information corresponding to each test question.

[0130] Optionally, the second acquiring unit is specifically configured to:

[0131] Obtaining an initial question bank corresponding to the teaching material to be evaluated;

[0132] The initial question bank is screened according to the teaching plan to be evaluated to obtain a target question bank.

[0133] Optionally, the first input unit is specifically used for:

[0134] Inputting a target test question in the target question bank into the target VS-GPT model to obtain a first output result corresponding to the target test question;

[0135] If the first output result corresponding to the target test question does not contain a terminator, inputting the first output result into the target VS-GPT model to obtain a second output result corresponding to the target test question;

[0136] If the second output result corresponding to the target test question includes an end symbol, the second output result is determined as the predicted solution information corresponding to the target test question.

[0137] Optionally, the training module 203 includes:

[0138] A third acquisition unit, used to acquire a word encoding vector, a position encoding vector, and a preset learnable encoding vector corresponding to the sample in the target sample set;

[0139] A second input unit is used to input the word encoding vector, the position encoding vector and the preset learnable encoding vector corresponding to the sample into the initial VS-GPT model to obtain a predicted mask text, wherein the predicted mask text includes: at least one predicted mask word;

[0140] A training unit is used to train the initial VS-GPT model according to an objective function formed by the predicted mask text and the mask text corresponding to the sample to obtain a target VS-GPT model.

[0141] Optionally, the initial VS-GPT model includes: at least two tansformer layers and at least one fully connected layer.

[0142] The above-mentioned product can execute the teaching data evaluation method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the teaching data evaluation method.

[0143] Embodiment 3

[0144] Figure 4 A schematic diagram of the structure of an electronic device 30 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0145] like Figure 4As shown, the electronic device 30 includes at least one processor 31, and a memory connected to the at least one processor 31, such as a read-only memory (ROM) 32, a random access memory (RAM) 33, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 31 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 32 or the computer program loaded from the storage unit 38 to the random access memory (RAM) 33. In the RAM 33, various programs and data required for the operation of the electronic device 30 can also be stored. The processor 31, the ROM 32, and the RAM 33 are connected to each other through a bus 34. An input / output (I / O) interface 35 is also connected to the bus 34.

[0146] A number of components in the electronic device 30 are connected to the I / O interface 35, including: an input unit 36, such as a keyboard, a mouse, etc.; an output unit 37, such as various types of displays, speakers, etc.; a storage unit 38, such as a disk, an optical disk, etc.; and a communication unit 39, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 39 allows the electronic device 30 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0147] The processor 31 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 31 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 31 performs the various methods and processes described above, such as the teaching data evaluation method:

[0148] Obtain the teaching data to be evaluated;

[0149] Determine a target sample set according to the teaching data to be evaluated;

[0150] Training an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model;

[0151] The target VS-GPT model is evaluated to obtain an evaluation result corresponding to the teaching data to be evaluated.

[0152] In some embodiments, the teaching data evaluation method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 38. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 30 via the ROM 32 and / or the communication unit 39. When the computer program is loaded into the RAM 33 and executed by the processor 31, one or more steps of the teaching data evaluation method described above can be performed. Alternatively, in other embodiments, the processor 31 can be configured to execute the teaching data evaluation method in any other appropriate manner (for example, by means of firmware).

[0153] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0154] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0155] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0156] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0157] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0158] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0159] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0160] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A teaching data evaluation method, characterized in that: include: Obtain the teaching data to be evaluated; Determine a target sample set according to the teaching data to be evaluated; Training an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model; The target VS-GPT model is evaluated to obtain an evaluation result corresponding to the teaching data to be evaluated.

2. The method according to claim 1, characterized in that: Determining a target sample set according to the teaching data to be evaluated includes: Acquire teaching data to be evaluated, wherein the teaching data to be evaluated includes: teaching materials to be evaluated and teaching plans to be evaluated; The teaching materials to be evaluated are sampled according to the teaching plan to be evaluated to obtain a target sample set.

3. The method according to claim 2, characterized in that The teaching materials to be evaluated are sampled according to the teaching plan to be evaluated to obtain a target sample set, including: Determine the proportion information and / or sorting information of each type of teaching materials to be evaluated according to the teaching plan to be evaluated; Each type of teaching material to be evaluated is sampled according to the proportion information and / or ranking information of each type of teaching material to be evaluated to obtain a target sample set.

4. The method according to claim 2, characterized in that: The target VS-GPT model is evaluated to obtain evaluation results corresponding to the teaching data to be evaluated, including: Obtaining a target question bank corresponding to the teaching data to be evaluated; Input the test questions in the target question bank into the target VS-GPT model in sequence to obtain predicted solution information corresponding to each test question, wherein the predicted solution information includes: at least one word; The evaluation result corresponding to the teaching data to be evaluated is determined according to the solution information corresponding to each test question and the predicted solution information corresponding to each test question.

5. The method according to claim 4, characterized in that Obtaining a target question bank corresponding to the teaching data to be evaluated includes: Obtaining an initial question bank corresponding to the teaching material to be evaluated; The initial question bank is screened according to the teaching plan to be evaluated to obtain a target question bank.

6. The method according to claim 4, characterized in that Input the test questions in the target question bank into the target VS-GPT model in sequence to obtain the predicted solution information corresponding to each test question, including: Inputting a target test question in the target question bank into the target VS-GPT model to obtain a first output result corresponding to the target test question; If the first output result corresponding to the target test question does not contain a terminator, inputting the first output result into the target VS-GPT model to obtain a second output result corresponding to the target test question; If the second output result corresponding to the target test question includes an end symbol, the second output result is determined as the predicted solution information corresponding to the target test question.

7. The method according to claim 1, characterized in that Training an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model includes: Obtaining a word encoding vector, a position encoding vector, and a preset learnable encoding vector corresponding to the samples in the target sample set; Inputting the word encoding vector, the position encoding vector and the preset learnable encoding vector corresponding to the sample into the initial VS-GPT model to obtain a predicted mask text, wherein the predicted mask text includes: at least one predicted mask word; The initial VS-GPT model is trained according to an objective function formed by the predicted mask text and the mask text corresponding to the sample to obtain a target VS-GPT model.

8. The method according to claim 7, characterized in that The initial VS-GPT model includes: at least two tansformer layers and at least one fully connected layer.

9. A teaching data evaluation device, characterized in that: include: An acquisition module is used to obtain the teaching data to be evaluated; A determination module, used to determine a target sample set according to the teaching data to be evaluated; A training module, used to train an initial VS-GPT model according to the target sample set to obtain a target VS-GPT model; The evaluation module is used to evaluate the target VS-GPT model and obtain the evaluation results corresponding to the teaching data to be evaluated.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the teaching data evaluation method described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the teaching data evaluation method described in any one of claims 1 to 8 when executed.

Images