CN118133801A - GPT-based multi-mode instruction generation and planning execution system and method - Google Patents
GPT-based multi-mode instruction generation and planning execution system and method Download PDFInfo
- Publication number
- CN118133801A CN118133801A CN202311838139.9A CN202311838139A CN118133801A CN 118133801 A CN118133801 A CN 118133801A CN 202311838139 A CN202311838139 A CN 202311838139A CN 118133801 A CN118133801 A CN 118133801A
- Authority
- CN
- China
- Prior art keywords
- instruction
- execution
- data
- gpt
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000008569 process Effects 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 7
- 238000012937 correction Methods 0.000 claims description 18
- 230000008859 change Effects 0.000 claims description 9
- 230000009471 action Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000001427 coherent effect Effects 0.000 claims description 4
- 238000013499 data model Methods 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000003058 natural language processing Methods 0.000 abstract description 3
- 238000007405 data analysis Methods 0.000 abstract description 2
- 238000012217 deletion Methods 0.000 abstract description 2
- 230000037430 deletion Effects 0.000 abstract description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The invention discloses a system and a method for generating and planning execution of a multi-mode instruction based on GPT, comprising the following steps: the invention relates to the technical field of natural language processing, in particular to a method for acquiring mass data and inputting the mass data into a value target object. According to the system and the method for generating and planning execution of the multi-modal instructions based on the GPT, the instructions are generated through natural voice texts in an instruction library in an input mode, the input instructions are analyzed, the analyzed deviation is corrected, the problem that the instructions are generated in error or cannot be generated due to the partial deletion of word sounds, fonts and word senses before the instructions are generated can be reduced, the instructions which do not occur are predicted after the instructions are generated through stored execution data analysis, and the GPT technology can generate natural language texts with identical thinking capability through training of massive data, so that the process of generating and planning execution of the multi-modal instructions is more accurate.
Description
Technical Field
The invention relates to the technical field of natural language processing, in particular to a system and a method for generating and planning execution of a multi-mode instruction based on GPT.
Background
GPT is a model for natural language processing, is usually used for generating human language text, can be understood as an artificial intelligent text interaction engine, is trained by massive corpuses, and can be understood and inferred according to input problems and rules of users and knowledge enriched in the corpuses, so that content required by the users is output.
In the existing multi-mode instruction generation and execution process, the problem that the instruction is generated incorrectly or cannot be generated due to the local deficiency of word voice, word shape and word meaning is lower in accuracy, the instruction and execution structure is relatively mechanized and is operated completely according to a set template, learning capacity is lacked in the multi-mode instruction generation and execution process, and limitation exists in the application process.
Disclosure of Invention
(One) solving the technical problems
In order to overcome the defects in the prior art, the invention provides a system and a method for generating and planning execution of a multi-mode instruction based on GPT, which solve the problems mentioned in the background art.
(II) technical scheme
In order to achieve the above purpose, the invention is realized by the following technical scheme: a method for generating and planning execution of a multi-mode instruction based on GPT specifically comprises the following steps:
Step one: acquiring mass data, inputting the mass data into a target object, training the mass data, generating a coherent natural voice text, generating an instruction library in the target object, and generating an instruction for the natural voice text in the instruction library in an input mode, wherein the target object is a GPT data model;
Step two: before an instruction is generated by natural voice text in an instruction library, the instruction in an input form is required to be analyzed, and the analyzed deviation is corrected, wherein the unmatched instruction information is as follows: instruction information which cannot be identified in the instruction library;
step three: after the instruction is generated, executing according to the instruction text, storing the executed instruction, analyzing the stored execution data, performing execution prediction on the instruction which does not occur, and prompting the predicted result, wherein the prompting execution information does not influence the forced execution of the instruction;
Step four: and optimizing the instruction information and the execution information to generate optimized instruction information and execution information.
As an improved technical scheme, in the first step, an instruction library is generated in a target object, and the specific mode of generating an instruction for natural voice text in the instruction library in an input mode is as follows:
s1: the method comprises the steps of sequentially marking mass data as A j according to data classification, wherein j=1, 2, & gt, and n, wherein n is the serial number of a data variety, sequentially marking data of the same kind as A jBj according to a consecutive sequence, and referring to a calculation formula: In the process of inputting the mass data A j, the data A jBj are in unordered arrangement;
S2: then classifying according to the data A jBj, wherein the data A jBj is a generalized unit and marked as GA j, a plurality of instruction representing units are generated in the GA j and marked as DGA j, input instruction information is marked as Sj, and when Sj is smaller than the DGA j, corresponding content is extracted in the DGA j and execution instruction information is generated.
As an improved technical scheme, the specific way of analyzing the input form of the instruction and correcting the analyzed deviation in the step two is as follows:
P1: firstly, setting a correction unit in the DGA j, marking the correction unit as alpha, inputting an input instruction Sj, and when the instruction representing unit GA j in the DGA j cannot identify the instruction information Sj, classifying the instruction information Sj under the action of the correction unit alpha, wherein the correction range of the correction unit alpha comprises: any one or combination of word sound, word shape and word sense;
P2: when the correction unit α cannot classify the instruction information Sj, the number of words in Sj is denoted as D s, where s=1, 2, & gt, n, s denote the number of words to be recognized, and the number of words to be recognized is sequentially recognized from more to less, and the highest value of the search amount is denoted as Sj max in D s, where Sj max∈DGAj.
As an improved technical scheme, when the sequences in the P2 from more to less are identified, the specific method is as follows:
p21: the search amount of D s is labeled PU n, the quantity threshold duty cycle is set, and labeled ST, according to the calculation formula: And when the K is larger than or equal to ST, the number of the words is not recognized in the order of more words, and the instruction for recognizing the current number of words is taken.
As an improved technical scheme, in the third step, the specific modes of executing and predicting the non-occurring instruction after analyzing the stored execution data and prompting the predicted result are as follows:
after the execution information is acquired, recording the execution information of each time, recording all the execution information in the time T according to the time sequence, generating a change curve graph in the time T, and predicting the result of the next execution information according to the change trend of the change curve graph.
As an improved technical solution, the specific way of optimizing the instruction information and the execution information in the fourth step is as follows: learning is performed on the basis of the original data, the disadvantages of the original data are analyzed, and the learning is performed on the basis of the disadvantages.
The invention also provides a computer readable storage medium, wherein a program is stored on the computer readable storage medium, and the program is executed by a processor to realize the method for generating and planning execution of the GPT-based multi-mode instruction.
The invention also provides an electronic device, which comprises at least one processor, and at least one memory and a bus which are connected with the processor; the processor and the memory complete communication with each other through the bus; the processor is used for calling program instructions in the memory to execute the GPT-based multi-mode instruction generation and execution planning method.
(III) beneficial effects
The invention provides a system and a method for generating and planning execution of a multi-mode instruction based on GPT. Compared with the prior art, the method has the following beneficial effects:
According to the system and the method for generating the multi-modal instruction and planning execution based on the GPT, the instruction is generated through natural voice texts in an instruction library in an input mode, the input mode instruction is analyzed, the analyzed deviation is corrected, the problem that the instruction is generated wrongly or cannot be generated due to the partial deletion of word sounds, fonts and word senses before the instruction is generated can be reduced, the non-generated instruction is executed and predicted after the instruction is generated through stored execution data analysis, and the GPT technology can generate the natural language texts with the same consistency of thinking capability through training massive data, so that the multi-modal instruction generation and planning execution process is more accurate.
Drawings
FIG. 1 is a flow chart of a method for generating and planning execution of GPT-based multi-modal instructions in accordance with the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the embodiment of the present invention provides five technical solutions:
Embodiment one: a method for generating and planning execution of a multi-mode instruction based on GPT specifically comprises the following steps:
Step one: acquiring mass data, inputting the mass data into a target object, training the mass data, generating a coherent natural voice text, generating an instruction library in the target object, and generating an instruction for the natural voice text in the instruction library in an input mode, wherein the target object is a GPT data model; in the first step, an instruction library is generated in a target object, and the specific mode of generating instructions for natural voice texts in the instruction library in an input mode is as follows: s1: the method comprises the steps of sequentially marking mass data as A j according to data classification, wherein j=1, 2, & gt, and n, wherein n is the serial number of a data variety, sequentially marking data of the same kind as A jBj according to a consecutive sequence, and referring to a calculation formula: In the process of inputting the mass data A j, the data A jBj are in unordered arrangement; s2: then classifying according to the data A jBj, wherein the data A jBj is a generalized unit and marked as GA j, a plurality of instruction representing units are generated in the GA j and marked as DGA j, input instruction information is marked as Sj, and when Sj is smaller than the DGA j, corresponding content is extracted in the DGA j and execution instruction information is generated.
Step two: before an instruction is generated by natural voice text in an instruction library, the instruction in an input form is required to be analyzed, and the analyzed deviation is corrected, wherein the unmatched instruction information is as follows: instruction information which cannot be identified in the instruction library; in the second step, the input form instruction is analyzed, and the specific mode of correcting the analyzed deviation is as follows: p1: firstly, setting a correction unit in the DGA j, marking the correction unit as alpha, inputting an input instruction Sj, and when the instruction representing unit GA j in the DGA j cannot identify the instruction information Sj, classifying the instruction information Sj under the action of the correction unit alpha, wherein the correction range of the correction unit alpha comprises: any one or combination of word sound, word shape and word sense; p2: when the correction unit α cannot classify the instruction information Sj, the number of words in Sj is denoted as D s, where s=1, 2, & gt, n, s denote the number of words to be recognized, and the number of words to be recognized is sequentially recognized from more to less, and the highest value of the search amount is denoted as Sj max in D s, where Sj max∈DGAj. When the identification is carried out from more to less in P2, the specific modes are as follows: p21: the search amount of D s is labeled PU n, the quantity threshold duty cycle is set, and labeled ST, according to the calculation formula: And when the K is larger than or equal to ST, the number of the words is not recognized in the order of more words, and the instruction for recognizing the current number of words is taken.
Embodiment two: the first embodiment is different from the first embodiment in that the third step: after the instruction is generated, executing according to the instruction text, storing the executed instruction, analyzing the stored execution data, performing execution prediction on the instruction which does not occur, and prompting the predicted result, wherein the prompting execution information does not influence the forced execution of the instruction; in the third step, the stored execution data is used for executing prediction on the non-generated instruction after analysis, and the specific mode for prompting the predicted result is as follows: after the execution information is acquired, recording the execution information of each time, recording all the execution information in the time T according to the time sequence, generating a change curve graph in the time T, and predicting the result of the next execution information according to the change trend of the change curve graph.
Step four: and optimizing the instruction information and the execution information to generate optimized instruction information and execution information. In the fourth step, the specific mode of optimizing the instruction information and the execution information is as follows: learning is performed on the basis of the original data, the disadvantages of the original data are analyzed, and the learning is performed on the basis of the disadvantages.
Embodiment III: the invention also provides a computer readable storage medium having stored thereon a program which when executed by a processor implements a method for GPT-based multimodal instruction generation and programming execution.
By generating instructions in the instruction library through natural voice texts in an input form, analyzing the instructions in the input form and correcting the analyzed deviation, the problem that the instructions are generated wrongly or cannot be generated due to the local deficiency of word sounds, fonts and word senses can be reduced before the instructions are generated, and after the instructions are generated, the instructions which do not occur are predicted after the stored execution data are analyzed, and the GPT technology can generate natural language texts with identical thinking capability through training mass data, so that the multi-modal instruction generation and planning execution processes are more accurate.
The invention also provides an electronic device, which comprises at least one processor, and at least one memory and a bus connected with the processor; the processor and the memory complete communication with each other through a bus; the processor is used for calling program instructions in the memory to execute the GPT-based multi-mode instruction generation and programming execution method.
Embodiment four: compared with the first, second and third embodiments, the technical solution of the present embodiment is to combine and implement the solutions of the first, second and third embodiments.
Fifth embodiment: a method for generating and planning execution of a multi-mode instruction based on GPT specifically comprises the following steps:
Step one: acquiring mass data, inputting the mass data into a target object, training the mass data, generating a coherent natural voice text, generating an instruction library in the target object, and generating an instruction for the natural voice text in the instruction library in an input mode, wherein the target object is a GPT data model;
Step two: before an instruction is generated by natural voice text in an instruction library, the instruction in an input form is required to be analyzed, and the analyzed deviation is corrected, wherein the unmatched instruction information is as follows: instruction information which cannot be identified in the instruction library;
step three: after the instruction is generated, executing according to the instruction text, storing the executed instruction, analyzing the stored execution data, performing execution prediction on the instruction which does not occur, and prompting the predicted result, wherein the prompting execution information does not influence the forced execution of the instruction;
Step four: and optimizing the instruction information and the execution information to generate optimized instruction information and execution information.
And all that is not described in detail in this specification is well known to those skilled in the art.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. A GPT-based method for generating and planning execution of multi-modal instructions, comprising: the method specifically comprises the following steps:
Step one: acquiring mass data, inputting the mass data into a target object, training the mass data, generating a coherent natural voice text, generating an instruction library in the target object, and generating an instruction for the natural voice text in the instruction library in an input mode, wherein the target object is a GPT data model;
Step two: before an instruction is generated by natural voice text in an instruction library, the instruction in an input form is required to be analyzed, and the analyzed deviation is corrected, wherein the unmatched instruction information is as follows: instruction information which cannot be identified in the instruction library;
step three: after the instruction is generated, executing according to the instruction text, storing the executed instruction, analyzing the stored execution data, performing execution prediction on the instruction which does not occur, and prompting the predicted result, wherein the prompting execution information does not influence the forced execution of the instruction;
Step four: and optimizing the instruction information and the execution information to generate optimized instruction information and execution information.
2. The GPT-based method of multimodal instruction generation and execution planning of claim 1, wherein: in the first step, an instruction library is generated in a target object, and the specific mode of generating instructions for natural voice texts in the instruction library in an input mode is as follows:
s1: the method comprises the steps of sequentially marking mass data as A j according to data classification, wherein j=1, 2, & gt, and n, wherein n is the serial number of a data variety, sequentially marking data of the same kind as A jBj according to a consecutive sequence, and referring to a calculation formula: In the process of inputting the mass data A j, the data A jBj are in unordered arrangement;
S2: then classifying according to the data A jBj, wherein the data A jBj is a generalized unit and marked as GA j, a plurality of instruction representing units are generated in the GA j and marked as DGA j, input instruction information is marked as Sj, and when Sj is smaller than the DGA j, corresponding content is extracted in the DGA j and execution instruction information is generated.
3. The GPT-based method of multimodal instruction generation and execution planning of claim 2, wherein: in the second step, the specific ways of analyzing the input form of the instruction and correcting the analyzed deviation are as follows:
P1: firstly, setting a correction unit in the DGA j, marking the correction unit as alpha, inputting an input instruction Sj, and when the instruction representing unit GA j in the DGA j cannot identify the instruction information Sj, classifying the instruction information Sj under the action of the correction unit alpha, wherein the correction range of the correction unit alpha comprises: any one or combination of word sound, word shape and word sense;
P2: when the correction unit α cannot classify the instruction information Sj, the number of words in Sj is denoted as D s, where s=1, 2, & gt, n, s denote the number of words to be recognized, and the number of words to be recognized is sequentially recognized from more to less, and the highest value of the search amount is denoted as Sj max in D s, where Sj max∈DGAj.
4. A method of GPT-based multimodal instruction generation and execution planning as in claim 3 wherein: when the sequence from more to less in the P2 is identified, the specific mode is as follows:
p21: the search amount of D s is labeled PU n, the quantity threshold duty cycle is set, and labeled ST, according to the calculation formula: And when the K is larger than or equal to ST, the number of the words is not recognized in the order of more words, and the instruction for recognizing the current number of words is taken.
5. The GPT-based method of multimodal instruction generation and execution planning of claim 1, wherein: in the third step, the stored execution data is used for executing prediction on the non-generated instruction after analysis, and the specific modes of prompting the predicted result are as follows:
after the execution information is acquired, recording the execution information of each time, recording all the execution information in the time T according to the time sequence, generating a change curve graph in the time T, and predicting the result of the next execution information according to the change trend of the change curve graph.
6. The GPT-based method of multimodal instruction generation and execution planning of claim 1, wherein: in the fourth step, the specific mode of optimizing the instruction information and the execution information is as follows: learning is performed on the basis of the original data, the disadvantages of the original data are analyzed, and the learning is performed on the basis of the disadvantages.
7. A computer readable storage medium having stored thereon a program, which when executed by a processor implements the GPT-based method of multimodal instruction generation and planning execution as claimed in any one of claims 1 to 6.
8. An electronic device comprising at least one processor, and at least one memory, bus connected to the processor; the processor and the memory complete communication with each other through the bus; the processor is configured to invoke program instructions in the memory to perform the GPT-based multi-modal instruction generation and execution planning method of any of claims 1 to 6.
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118133801A true CN118133801A (en) | 2024-06-04 |
Family
ID=
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110442859B (en) | Labeling corpus generation method, device, equipment and storage medium | |
| CN109800414A (en) | Faulty wording corrects recommended method and system | |
| CN113987179A (en) | Knowledge enhancement and backtracking loss-based conversational emotion recognition network model, construction method, electronic device and storage medium | |
| CN112528637A (en) | Text processing model training method and device, computer equipment and storage medium | |
| CN113657098B (en) | Text error correction method, device, equipment and storage medium | |
| CN109522550A (en) | Text information error correction method, device, computer equipment and storage medium | |
| CN111428470A (en) | Text continuity judgment method, text continuity judgment model training method, electronic device and readable medium | |
| CN106649210A (en) | Data conversion method and device | |
| CN113012701B (en) | Identification method, identification device, electronic equipment and storage medium | |
| CN109726386B (en) | Word vector model generation method, device and computer readable storage medium | |
| CN113626563A (en) | Method and electronic equipment for training natural language processing model and natural language processing | |
| CN117112767A (en) | Question and answer result generation method, commercial query big model training method and device | |
| CN115858776B (en) | Variant text classification recognition method, system, storage medium and electronic equipment | |
| CN111400340A (en) | Natural language processing method and device, computer equipment and storage medium | |
| CN116186223A (en) | Financial text processing method, device, equipment and storage medium | |
| CN115658921A (en) | Open domain scientific knowledge discovery method and device based on pre-training language model | |
| CN118133801A (en) | GPT-based multi-mode instruction generation and planning execution system and method | |
| CN115600595A (en) | Entity relationship extraction method, system, equipment and readable storage medium | |
| CN115587163A (en) | Text classification method and device, electronic equipment and storage medium | |
| CN114822726A (en) | Construction method, analysis method, device, storage medium and computer equipment | |
| CN115358473A (en) | Power load prediction method and prediction system based on deep learning | |
| CN110413779B (en) | Word vector training method, system and medium for power industry | |
| CN114896966A (en) | Method, system, equipment and medium for positioning grammar error of Chinese text | |
| CN114741512A (en) | Automatic text classification method and system | |
| KR20220073644A (en) | Question answering system by using constraints and information provision method thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication |