40 results about "Model translation" patented technology

The invention belongs to the technical field of text translation, and provides a business text real-time translation method and system based on artificial intelligence, and the method comprises the steps: intelligent agent construction, to-be-translated text data collection, scene classification probability data calculation, translation subject model matching, text data translation and feedback information processing. According to the invention, through a multi-agent collaborative architecture, a translation task of a specific scene is realized, and the applicability of translation is improved; through a TF-IDF algorithm and a BERT model, dynamic scheduling and accurate matching of intelligent agents are realized, and the reliability of a translation model is improved; by integrating loss functions of overall loss, semantic loss, literary and cultural loss, the sensitivity of the translation model in terms of semantics, literary and cultural aspects is improved, and the model translation accuracy and peer-to-peer are improved; the user experience is improved and translation styles are switched by collecting user feedback and replacing the intelligent agent.

The invention relates to a multi-language large model translation performance optimization method and system based on prompt strategy activation, and belongs to the field of natural language processing. In a low-resource language environment, the translation performance of a multi-language large model still has insufficiency. The understanding ability of a large model on low-resource language input semantics is limited, so that part of potential translation ability of the large model is not fully activated, in order to relieve the influence, prompts are screened from a high-quality corpus based on four prompt strategies, a context learning template is constructed to serve as large model user input, and therefore the user input of the large model is improved. Task guidance serves as system input, and finally performance optimization of a large model in a low-resource language translation task is achieved by dynamically adjusting a prompt strategy and model parameters. According to the method, the potential translation ability of a large model can be effectively activated, and the translation quality of the large model in a low-resource language environment is remarkably improved.

The invention relates to a training method of a translation model, a text translation method and a training method of a generative model. The method comprises the steps that a training text of a first language is input into an initial translation model, a model output text is obtained, and the model output text comprises a model translation text of a second language; obtaining a format reward corresponding to the model output text based on the format of the model output text, wherein the format reward is used for representing whether the model output text accords with a preset format; obtaining a measurement reward corresponding to the model translation text based on the training text and the model translation text, wherein the measurement reward is used for representing the translation quality of the model translation text; obtaining a target reward according to the format reward and the measurement reward; and training the initial translation model based on the target reward and a preset reinforcement learning algorithm to obtain a target translation model. According to the training method, a format and measurement mixed reward mode is adopted, a guidance signal with rich and effective information is provided for reinforcement learning, and the translation quality of the translation model can be improved.

A system is provided.SOLUTION: A system, comprising: means for collecting audio; means for pre-processing the collected audio; means for converting the pre-processed audio to text using a AI recognition model; means for translating the converted text using a multilingual generative AI model; means for displaying and playing the translated text; means for expert review and completion of the translated text; means for data encryption; and means for training the generative speech model to account for cultural differences.SELECTED DRAWING: Figure 1

The invention provides a large model context learning machine translation method based on word granularity alignment, which relates to the field of natural language processing, and comprises the following steps: an external knowledge assisting stage: performing multi-level retrieval matching on a source text word alignment set; a large model translation stage: obtaining a large model translation set, and taking the large model translation set as one of candidate translations; in the post-selection stage, the obtained source text word alignment set, the obtained external dictionary word alignment set, the obtained entity library alignment set, the obtained embedding representation of the auxiliary translation set and the obtained large model translation set are subjected to similarity calculation scoring for multiple times, and screening is conducted according to similarity scores to obtain a candidate word alignment set; a prompt template is designed according to a task, a source text word alignment set and a candidate word alignment set are put into the prompt template, and a highly-aligned external word alignment set is used for large model context learning to generate an optimal translation result; according to the method, various translation errors of a large model in a low-resource environment are relieved.

The invention relates to the field of computers, and provides a Modelica model state curing method, device, equipment, medium and product, and the method comprises the steps: responding to a translation simulation request of a to-be-cured model, translating the to-be-cured model, and obtaining model uncertainty representation, the model uncertainty representation comprises a state variable selected by translation, an initial value of an initial supplement variable, a nonlinear iteration variable and an iteration initial value of the nonlinear iteration variable, and the initial supplement variable refers to a variable needing to be supplemented with the initial value; and executing at least one of the following steps and finally writing into the model so as to carry out state solidification on the translated model: determining the choice of the state variable, locking the certainty of the initial value, providing the iteration initial value of the nonlinear iteration variable and marking the nonlinear iteration variable by utilizing the manufacturer annotation. According to the method, the device, the equipment, the medium and the product provided by the invention, ambiguity in a model translation process is eliminated, and high consistency and reliability of cross-platform simulation results are realized.

The invention discloses a multi-language internationalization translation method and system, a medium and equipment, and is applied to a distributed control system.The method comprises the steps that a background processing program reads a user visible text wrapped by an internationalization identifier in a source file and stores the user visible text as a json file; the automatic translation large model translates the user visible text to obtain the translated user visible text, and the background processing program stores the translated user visible text into the json file again; when the distributed control system runs based on a source code, automatically loading a dynamic library, matching a user visible text required to be translated of an internationalized identifier package with a json file, and displaying the user visible text required to be translated as a translated user visible text based on a result returned after matching, the multi-language version can be obtained without modifying a large number of source codes, the system language switching efficiency is improved, the labor cost is reduced, and the system maintenance efficiency is improved.

A real time translation method for a game includes extracting features from video frames using computer vision and a database, performing an association process to find a machine learning model best matching the features for translation, obtaining texts in the game through optical character recognition (OCR), preprocessing the texts, translating the texts using the machine learning model to generate translated texts, and rendering the translated texts to images of the video frames for displaying the images with the translated texts on a display device.

The invention provides a cloud analysis data asset classification and level-to-level management system, and relates to the technical field of asset management. The cloud analysis data asset classification and level-to-level management system comprises a data acquisition layer, the data acquisition layer is connected with a metadata extraction layer and a preprocessing layer, the preprocessing layer is connected with a classification and level-to-level engine layer, the classification and level-to-level engine layer is connected with a multi-language AI large model translation layer and then connected with a storage layer, and the storage layer is connected with an application layer. The metadata extraction layer is a metadata extraction algorithm based on deep learning and an attention mechanism and can automatically identify and extract key metadata fields from original data and keep metadata unchanged, and the preprocessing layer cleans, converts and standardizes the collected data. By combining metadata extraction, AI large model translation and classification and grading technologies, the automation level of data asset management is improved, and efficient management and protection of data assets are achieved.

The invention discloses a building carbon emission decision-making method, device and equipment based on a full life cycle, a medium and a product. The method comprises the following steps: acquiring a plurality of design schemes and corresponding morphological parameters of a large-space public building; performing parametric modeling and model translation on the morphological parameters to obtain a morphological description model and a substance description model; calculating year-by-year carbon emission data of each design scheme in the full life cycle of the large-space public building according to energy consumption data of the form description model and material data and carbon emission factors of the material description model; according to the year-by-year carbon emission data of each design scheme, calculating the weight of each year-by-year life cycle stage of each design scheme, and obtaining a weighting matrix; and sorting the design schemes according to the weighting matrix, and selecting an optimal design scheme according to a sorting result. By adopting the embodiment of the invention, the carbon emission calculation can be carried out in the early stage of building design, and the scheme with the optimal carbon emission performance can be evaluated and decided, so that the construction of energy-saving and carbon-reducing buildings is facilitated.

The invention provides a model training method, a translation method, electronic equipment and a computer program product, and relates to the technical field of interface design. The model training method comprises the following steps: constructing an initial model for language translation; extracting visual features of various fonts in the font database; based on historical region structure parameters of a historical region in the historical design interface, determining semantic constraint conditions for constraining the length of the historical translation text; and performing joint training on the initial model according to the visual features and the semantic constraint conditions to obtain a constraint translation model. The translation method comprises the following steps: determining initial text information and regional structure parameters according to a text region in a design interface; processing according to the initial text information and the regional structure parameters through a constraint translation model to obtain a translated text; and filling each translated text into the corresponding text region.

The application provides a knowledge learning method and device of a machine translation model, comprising: constructing a migration dataset; respectively using a machine translation teacher model and a machine translation student model to translate the original text of each sample of the migration dataset to obtain the first candidate word probability distribution and the second candidate word probability distribution of each sample of the migration dataset; determining the migration loss of each sample of the migration dataset according to the first candidate word probability distribution, the second candidate word probability distribution and the standard candidate word probability distribution of each sample of the migration dataset; and realizing the knowledge learning of the machine translation student model to the machine translation teacher model based on the first candidate word probability distribution of each sample of the migration dataset and the migration loss. The application enables the machine translation student model to learn complementary knowledge from the machine translation teacher model through the migration loss, realizes the knowledge accumulation of the machine translation student model, and further improves the translation performance of the machine translation student model.

The invention discloses a real-time context sensing dynamic cue word generation method based on large model translation. The method comprises the steps of initialization, context sensing and core concept extraction, dynamic cue word generation, new concept recognition and list updating. According to the method, a traditional static translation process is converted into a dynamic process with memory and state maintenance capabilities, a global core concept list is constructed and dynamically maintained in real time, and dynamic cue words containing determined standard translations are automatically generated in the translation process; the global consistency of the translations on core concepts such as term named entities is remarkably improved, the problem that core concept translations are inconsistent in long document translation is fundamentally solved, and the method is suitable for professional fields such as technical documents, academic papers and commercial reports with extremely high term consistency requirements.

The invention provides an automatic error detection and repair method and system for translation from natural language to SQL (Structured Query Language), which is characterized in that a plurality of candidate SQL queries are generated for natural language queries for given natural language and original SQL queries on the basis of a correctness condition, namely execution consistency, matched with the natural language queries and the SQL queries; the method comprises the following steps of: firstly, translating a natural language to an SQL (Structured Query Language), trying to generate a counter example between non-equivalent SQL queries by using an SQL equivalence verification technology, respectively executing the natural language and the SQL queries on the counter example by using a large language model and a database execution engine, and if the execution result of the natural language is inconsistent with the execution result of the original SQL query, judging that an error exists in the translation of converting the natural language to the SQL; and if the execution result of the candidate SQL query on the counter example is more matched with the natural language, taking the candidate SQL as a repair result. By means of the method, wrong translation in an existing Text-to-SQL data set can be automatically detected and repaired, higher-quality training and testing data sets are provided, meanwhile, the method can be used for Text-to-SQL model translation tasks, and translation accuracy is improved.

The application discloses a kind of banking data processing method, system, terminal equipment and readable storage medium, it is related to banking data processing field, the banking data processing method of the present application includes: by uniform syntax specification, generate the database modeling requirement information of multiple different databases;According to the database modeling requirement information of each database, and in combination with uniform logic modeling structure, establish banking professional knowledge base;Acquire the banking data to be processed, by banking professional knowledge base, the data logic model corresponding to banking data is translated into the data physical model corresponding to banking data;According to data physical model, generate the modeling statement adapted to each database;Modeling statement is input to the database environment corresponding to each database and is executed, obtains the result of analyzing and processing banking data.Processing analysis efficiency of banking data can be improved and cost is reduced by using the present application.

The application relates to the computer field and provides a Modelica model state solidification method, device, equipment, medium and product, wherein the method comprises the following steps: in response to a translation simulation request of a to-be-solidified model, the to-be-solidified model is translated to obtain a model uncertainty representation, the model uncertainty representation comprises translation-selected state variables, initial values of complementary variables, nonlinear iteration variables and iteration initial values of the nonlinear iteration variables, and the complementary variables refer to variables that need to be supplemented with initial values; at least one of the following steps is executed and finally written into a model to solidify the state of the translated model: determining the necessity of the state variables, determining the certainty of the locked initial value, providing the iteration initial value of the nonlinear iteration variable and marking the nonlinear iteration variable by using a manufacturer annotation. The method, device, equipment, medium and product provided by the application eliminate the ambiguity in the model translation process, and realize the high consistency and reliability of cross-platform simulation results.

The invention relates to an image processing method and device, a storage medium and a program product, and belongs to the technical field of image processing, and the method comprises the steps: obtaining a character string in a target image, and obtaining a first character string; inputting the first character string into a recognition model to obtain a second character string which is output by the recognition model and obtained by translating the first character string; wherein the recognition model is obtained through training by taking training data as input and taking a translation result of a training character string in the training data as at least part of output, the training data comprises first sample data, and the first sample data is generated based on a sample image; the first sample data comprises a sample character string in the sample image and a contextual relationship of the sample character string in the sample image. In this way, the second character string obtained through translation of the recognition model has higher accuracy.

The application belongs to the technical field of text translation, and provides a business text real-time translation method and system based on artificial intelligence, which comprises the following steps: intelligent agent construction, acquisition of text data to be translated, calculation of scene classification probability data, matching of translation subject model, text data translation, and feedback information processing; through the multi-agent collaborative architecture, the application realizes the translation task in a specific scene and improves the applicability of translation; through the TF-IDF algorithm and the BERT model, the dynamic scheduling and accurate matching of the intelligent agent are realized, and the reliability of the translation model is improved; through the loss function integrating the overall loss, the semantic loss, the style loss and the cultural loss, the sensitivity of the translation model in terms of semantics, style and culture is improved, and the model translation accuracy and parity are improved; through the collection of user feedback and the replacement of intelligent agents, the improvement of user experience and the switching of translation styles are realized.

The invention discloses a multi-language voice real-time translation method and system based on a large model, and belongs to the technical field of voice real-time translation, and the system comprises a voice collection module which is used for pre-collecting a voice sample of a speaker and collecting to-be-translated voice in real time; the voice recognition module is used for converting voice to be translated into a source language text; the large model translation module is used for sending the source language text to a selected large model API through a network request and receiving a target language text obtained after the selected large model translates the source language text in real time; the timbre feature extraction module is used for carrying out timbre feature extraction on a pre-collected voice sample; and the speech synthesis and output module is used for generating target language speech about the speaker from the extracted timbre features and the target speech text through a speech synthesis model. According to the method, the user experience is enhanced, the end-to-end translation delay is reduced, and application scenes such as real-time dialogues and conferences with extremely high delay requirements are met.

This invention relates to a method for automatically constructing a large-scale Chinese-Myanmar parallel corpus based on multidimensional evaluation, belonging to the field of natural language processing technology. To address the scarcity of Chinese-Myanmar parallel corpus resources, this invention proposes a method for automatically constructing a large-scale Chinese-Myanmar parallel corpus based on multidimensional evaluation. First, Burmese corpora are manually collected to construct a basic dataset. Second, prompt words are designed to generate a candidate set of pseudo-parallel Chinese-Burmese corpora using a large-scale model. Then, the candidate set is filtered using semantic similarity constraints. Finally, the large-scale model is used to score the filtered candidate set from multiple dimensions, and the corpora are ranked and hierarchically filtered based on the scores to obtain high-quality pseudo-parallel corpora. Experimental results show that the corpus constructed by this invention can effectively improve the model's translation performance and significantly enhance its translation quality in low-resource Chinese-Myanmar language environments.

The embodiment of the invention provides a translation difficulty assessment method and device of a text translation model, and relates to the technical field of text processing. The method comprises the following steps: acquiring a source text vector from an encoder of a target text translation model, acquiring a translation text vector from a decoder, and performing global feature extraction on the source text vector and the translation text vector to obtain a corresponding global source vector and a corresponding global translation vector; and performing semantic fusion on the global source vector and the global translation vector to obtain a first semantic vector and a second semantic vector, obtaining a deep feature vector based on the first semantic vector and the second semantic vector, and inputting the deep feature vector into an output layer to perform difficulty score mapping to obtain a translation difficulty evaluation value corresponding to the target text translation model. Through a multi-level mining process, the overall feature mining capability of the target text translation model is obtained, and an accurate information basis is provided for difficulty assessment, so that the accuracy of difficulty assessment is improved.

The application provides a foreign language bill image translation method, device and medium based on artificial intelligence, and belongs to the technical field of data information. The foreign language bill image translation method comprises the following steps: acquiring a foreign language bill image, inputting the foreign language bill image into an image recognition model, and outputting bill text information and position coordinate information; inputting the bill text information into a pre-trained bill translation special model to obtain Chinese bill information; inputting the information into a large language model, optimizing the Chinese bill information according to a preset bill question and answer prompt word; using the position coordinate information, arranging and combining the optimized Chinese bill information according to a Chinese bill format, and drawing a Chinese bill image; using a key information extraction model to extract bill key information from the Chinese bill information; and displaying the Chinese bill image and the bill key information. The application can solve the problem that the prior art cannot accurately translate bill data sets, thereby causing difficulty in training accurate and effective bill information extraction.

This invention relates to a large-scale model translation fine-tuning method based on the importance awareness of entropy-varying parameters, belonging to the field of efficient large-scale model fine-tuning technology. Addressing the problem of poor model performance caused by data scarcity in low-resource tasks, this invention proposes a method that first freezes all trainable parameters, feeds the input data of the low-resource task into a pre-trained large language model, performs multiple forward propagations, and collects the entropy sequence of the output distribution of each Transformer layer. Based on the entropy sequence, the importance score of each layer is calculated from two dimensions: the intensity of inter-layer information change and the stability of intra-layer response. According to the importance score, a LoRA rank is dynamically assigned to each layer. The LoRA module with the assigned differential ranks is then activated, and the model undergoes standard fine-tuning training to obtain a fine-tuned model adapted to the target low-resource task. This invention effectively improves the performance of large models in low-resource translation tasks without increasing the number of trainable parameters.

To provide a system.SOLUTION: A system includes means for generating a button for sending a message in a specific language, means for inputting an input message as a prompt sentence to a generative AI model when the button is pressed, means for the generative AI model to translate the input message, and means for sending the translated message to a transmission destination.SELECTED DRAWING: Figure 1

This invention relates to an efficient fine-tuning method for translation parameters of large model based on hierarchical activation difference perception, belonging to the field of natural language processing and large model fine-tuning technology. To address the problems of pre-training bias dominance and hierarchical functional imbalance in existing large language models (LLMs) for low-resource machine translation tasks, this invention includes: firstly, accurately locating model modules sensitive to the machine translation task using cumulative activation intensity and cross-layer activation variance; then, dividing the model hierarchy into task-adaptive layers and legacy inert layers based on activation difference indices; introducing learnable latent gating parameters to generate hierarchical-specific modulation factors; and during efficient parameter fine-tuning, using the modulation factors to amplify the signal in the task-adaptive layer while suppressing the legacy inert layers. This invention effectively improves translation performance from Chinese to low-resource languages ​​such as Lao and Khmer by structurally differentiating the optimization direction.

The invention provides a document translation method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining a first document to be translated, wherein the first document comprises a first text and format information used for describing the first text; extracting the first text, and indicating a first model to translate the first text to obtain a first translated text; instructing the first model to reconstruct a first label based on the format information, and inserting the first label into the first translated text to obtain a translated document corresponding to the first document; and displaying the translated document corresponding to the first document. As a result, high-quality tagged document translation is achieved.

The invention provides a foreign language bill image translation method and device based on artificial intelligence and a medium, and belongs to the technical field of data information. The foreign language bill image translation method comprises the following steps: acquiring a foreign language bill image, inputting the foreign language bill image into an image recognition model, and outputting to obtain bill character information and position coordinate information; inputting the bill character information into a pre-trained bill translation special model, and translating to obtain Chinese bill information; inputting the information into a large language model, and optimizing the Chinese bill information according to a preset bill question and answer prompt word; the position coordinate information is used, the optimized Chinese bill information is arranged and combined according to a Chinese bill format, and a Chinese bill image is drawn; extracting bill key information from the Chinese bill information by using a key information extraction model; and displaying the Chinese bill image and the bill key information. The problem that effective bill information is difficult to train and accurately extract due to the fact that a bill data set is difficult to accurately translate in the prior art can be solved.

The invention provides a device, method and equipment for accelerating ISP simulation based on a large model and a storage medium, and the device comprises an obtaining module which is used for obtaining a preset code of each module in an ISP; the verification platform is used for processing the test chart by running the preset codes and preset parameters for each preset code to obtain a reference result; the translation module is used for translating a preset code based on a preset large model to obtain an initial code; the initial code runs on the GPU; the GPU is used for processing the test chart by running the initial code and preset parameters to obtain an initial result; the adjusting module is used for adjusting the initial code based on a preset large model to obtain a target code if the reference result is not matched with the initial result; the target result is matched with the reference result; the target result is obtained by operating the target code and the preset parameter processing test chart. According to the technical scheme, the initial code does not need to be compiled and adjusted manually, and therefore the simulation process of the ISP is accelerated.

The invention discloses an interactive multi-modal prompt fusion method for continuous learning of sign language translation, and aims to solve the problems of disastrous forgetting and insufficient multi-modal interaction consideration of an existing sign language translation system in continuous learning. According to the method, a fusion mechanism of task perception prompt and data perception prompt is constructed to realize balance of model stability and plasticity, cross-modal and cross-task interaction loss modules are constructed in training loss, information between modals and between tasks is utilized to prompt model translation accuracy, and meanwhile, low-rank decomposition is introduced to further reduce the number of parameters, so that the model translation accuracy is improved. And relatively light-weight fine tuning and learning are realized. The method improves the stability of continuous learning under the condition of guaranteeing the sign language translation accuracy, is suitable for a barrier-free communication scene of a hearing-impaired group, and is compatible with a mainstream multi-mode sign language translation framework.

To enable the execution of appropriate consensus building regarding translation content.SOLUTION: A server device 10 includes: an input control unit 13 which inputs voice data or text data received from a user terminal to a translation generation model with prompts set thereon for translating input voice data or text data into a designated language in a natural context; and a summary generation unit 15 which summarizes the contents of a conference in a designated language for each user, using translation data, which is translated voice data or text data output from the translation generation model in accordance with the input by the input control unit 13; and a consensus building unit 62 which receives an input of consensus or dissent from the user based on the summary generated by the summary generation unit 15.SELECTED DRAWING: Figure 34