DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-12 are pending and have been examined.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 11/17/2025 and 7/22/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement was considered and attached by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-12 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1 the limitations of “obtaining a prompt template and a generation content dependency relationship corresponding to each of human-machine dialogue elements in a target scenario, wherein the human-machine dialogue elements at least include: user queries and response content”, “progressively generating generation content of the corresponding human-machine dialogue elements based on the prompt templates and a pre-trained large language model according to the generation content dependency relationships”, and “generating multi-round dialogue data in the target scenario based on the generation content respectively corresponding to the user queries and the response content”, as drafted, are processes that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. More specifically, the mental process of a human thinking of a question and potential questions in the context of a scenario, thinking of a response to the question based on the scenario and a set of instructions or rules, and further thinking of additional questions and response to the question in the scenario. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the --Mental Processes-- grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
Regarding claim 10 the limitations of “obtaining multiple sets of multi-round dialogue data in a target scenario, wherein each set of the multi-round dialogue data comprises”, “multi-round user queries and response content arranged in a logically progressive sequence”, “constructing a dialogue content generation model based on a pre-trained large language model”, and “fine-tuning the dialogue content generation model based on the multiple sets of multi-round dialogue data until predicted loss value of the dialogue content generation model satisfies a preset convergence condition, wherein the predicted loss value is calculated based on single-round dialogue prediction loss of each set of multi-round dialogue data, and the single-round dialogue prediction loss is negative log-likelihood mean obtained by modeling predicted response content generated by the dialogue content generation model in response to the user queries in each round of dialogue data”, as drafted, are processes that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. More specifically, the mental process of a human reading a dialogue of questions and answers arranged logically, thinking of the answer vocabulary that was read in relation to the answer vocabulary that was thought of in the mind, and adjusting rules or instructions of how to answer the question in the mind. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the --Mental Processes-- grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
Regarding claim 11 the limitations of “in response to receiving a current-round user query, obtaining dialogue data from a specified number of previous rounds of dialogue”, “generating reply dialogue data to be responded to based on the current-round user query and the specified number of previous rounds of dialogue data, according to a logical progression”, “invoking a pre-trained dialogue content generation model based on the reply dialogue data to obtain response content output by the pre-trained dialogue content generation model”, and “responding to the current-round user query based on the response content”, as drafted, are processes that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. More specifically, the mental process of a human reading a dialogue of questions and answers in relation to the current question and thinking of the answer to the question according to rules or instructions in the mind. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the --Mental Processes-- grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application because there are no additional elements that integrate the judicial exception into a practical application. The claims are directed to an abstract idea.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The methods of the claim limitation can practically be performed in the mind or with the use of a pen or pencil.
With respect to claim 2, the claim recites “determining, according to the generation content dependency relationships, a first prompt template and a second prompt template from the prompt templates; generating generation content of the corresponding human-machine dialogue elements based on the first prompt template and the pre-trained large language model; progressively generating generation content of the corresponding human-machine dialogue elements based on the second prompt template, pre-generated target generation content, and the pre-trained large language model, according to the generation content dependency relationships”, which reads on a human thinking of a dialogue according to question templates. No additional limitations are present.
With respect to claim 3, the claim recites “respectively using the first prompt template as a current prompt template, and performing the following first content generation operations: generating a first current prompt based on the current prompt template; invoking the pre-trained large language model based on the first current prompt to generate the generation content of the corresponding human-machine dialogue elements”, which reads on a human thinking of a dialogue according to question templates. No additional limitations are present.
With respect to claim 4, the claim recites “determining, based on the generation content dependency relationships, a generation order of the generation content corresponding to the human-machine dialogue elements for the second prompt template; sequentially executing second content generation operations from first to last according to the generation order, wherein the second content generation operations comprise: obtaining, based on the generation content dependency relationships, pre-generated generation content on which the generation content of a current human-machine dialogue element depends, as the target generation content; formatting the second prompt template corresponding to the current human-machine dialogue element based on the target generation content, to generate a second current prompt; invoking the pre-trained large language model based on the second current prompt to generate the generation content of the current human-machine dialogue element”, which reads on a human thinking of a sequence of dialogue in question and answer format according to question templates. No additional limitations are present.
With respect to claim 5, the claim recites “wherein the generation content dependency relationships are represented by placeholders set in the prompt templates”, which reads on a human reading or identifying questions with placeholders. No additional limitations are present.
With respect to claim 6, the claim recites “wherein the prompt templates represent, through placeholders corresponding to a first human-machine dialogue element, that input conditions on which the generation of the generation content of a second human-machine dialogue element depends include the generation content of the first human-machine dialogue element, wherein the second human-machine dialogue element is the human-machine dialogue element corresponding to the prompt template”, which reads on a human determining a question from a prompt template and previous dialogue. No additional limitations are present.
With respect to claim 7, the claim recites “wherein the prompt template corresponding to the user queries includes a rule prompt, the rule prompt being used to instruct the pre-trained large language model to generate user queries according to a first rule, wherein the first rule includes: each group of generated user queries has relevance and a logical progression, and incorporates at least N instances of context, where N is a natural number greater than 1”, which reads on a human utilizing a question template that includes rules. No additional limitations are present.
With respect to claim 8, the claim recites “wherein the prompt template corresponding to the response content includes an instruction prompt, the instruction prompt being used to instruct the pre-trained large language model to generate the response content to a final-round user query in dialogue context based on prior dialogue context”, which reads on a human utilizing a question template that includes instructions. No additional limitations are present.
With respect to claim 9, the claim recites “wherein the user queries have a logical progression, and the response content corresponds to the user queries, the multi-round dialogue data in the target scenario being generated based on the generation content corresponding to the user queries and the response content, comprising: for each group of user queries, constructing an ordered combination of multi-round question-answer pairs corresponding to each group of user queries based on the user queries and the corresponding response content, according to the logical progression of the user queries; generating the multi-round dialogue data in the target scenario based on the ordered combination”, which reads on a human thinking of multi round dialogue that includes questions and answers. No additional limitations are present.
With respect to claim 12, the claim recites “wherein the pre-trained dialogue content generation model is trained by: obtaining multiple sets of multi-round dialogue data in a target scenario, wherein each set of the multi-round dialogue data comprises: multi-round user queries and response content arranged in a logically progressive sequence; constructing a dialogue content generation model based on a pre-trained large language model; fine-tuning the dialogue content generation model based on the multiple sets of multi-round dialogue data until predicted loss value of the dialogue content generation model satisfies a preset convergence condition, wherein the predicted loss value is calculated based on single-round dialogue prediction loss of each set of multi-round dialogue data, and the single-round dialogue prediction loss is negative log-likelihood mean obtained by modeling predicted response content generated by the dialogue content generation model in response to the user queries in each round of dialogue data”, which reads on a human thinking of multi round dialogue that includes questions and answers where the answers include predicted vocabulary. No additional limitations are present.
These claims further do not remedy the judicial exception being integrated into a practical application and further fail to include additional elements that are sufficient to amount to significantly more than the judicial exception.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-12 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Hettige et al. (U.S. PG Pub No. 20250094821), hereinafter Hettige.
Regarding claim 1 Hettige teaches:
A method for generating dialogue data, comprising: (P0033, The method can include accessing a set of training examples, wherein each training example of the set of training examples includes a dialog script between a user and a digital assistant, generating a set of synthesized training examples using an iterative process that is performed for each of one or more predefined scenarios.) obtaining a prompt template and a generation content dependency relationship corresponding to each of human-machine dialogue elements in a target scenario, wherein the human-machine dialogue elements at least include: user queries and response content; (P0033, Accessing a dialog script and corresponding prompt template and response template for a predefined scenario.; P0102, Simulated user interaction approaches involve generating dialogs mimicking how users might engage with a digital assistant. The approaches usually generate multi-turn, realistic conversations where users ask questions, make requests, or provide feedback. For example, simulating conversational flows allows a planner to script various user-initiated dialogues, such as booking a service, requesting recommendations, or asking for status updates.)
progressively generating generation content of the corresponding human-machine dialogue elements based on the prompt templates and a pre-trained large language model according to the generation content dependency relationships; (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM. This synthetic data can be created through a synthetic data generation pipeline, which is designed to populate a training dataset. The pipeline works by defining a series of dialogue scripts between a user and a digital assistant, complete with input/output templates and placeholders for specific slot values. These templates provide structured frameworks, allowing for a wide range of synthetic dialogues to be generated automatically or semi-automatically.)
generating multi-round dialogue data in the target scenario based on the generation content respectively corresponding to the user queries and the response content. (P0164, The synthetic data enhances the training data by adding secondary-happy-path data and non-sequitur data. For example, the synthetic data can be categorized into multi-action data (data involving in a scenario that relates to multiple actions, e.g., a conversation for finding restaurants and reserve a restaurant).)
Regarding claim 2 Hettige teaches claim 1 and further teaches:
wherein progressively generating the generation content of the corresponding human-machine dialogue elements based on the prompt templates and the pre-trained large language model according to the generation content dependency relationships comprises: determining, according to the generation content dependency relationships, a first prompt template and a second prompt template from the prompt templates; (P0095, The pipeline works by defining a series of dialogue scripts between a user and a digital assistant, complete with input/output templates and placeholders for specific slot values. These templates provide structured frameworks, allowing for a wide range of synthetic dialogues to be generated automatically or semi-automatically.)
generating generation content of the corresponding human-machine dialogue elements based on the first prompt template and the pre-trained large language model; (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM.)
progressively generating generation content of the corresponding human-machine dialogue elements based on the second prompt template, pre-generated target generation content, and the pre-trained large language model, according to the generation content dependency relationships. (P0073, The output data is appended to the utterance to construct an output prompt for input to the LLM.. … The context is retrievable from the context and memory store and includes user session information, dialog state, conversation or contextual history, user information, or any combination thereof. The LLM generates responses based on the output prompt.)
Regarding claim 3 Hettige teaches claim 2 and further teaches:
respectively using the first prompt template as a current prompt template, and performing the following first content generation operations: (P0095, The pipeline works by defining a series of dialogue scripts between a user and a digital assistant, complete with input/output templates and placeholders for specific slot values. These templates provide structured frameworks, allowing for a wide range of synthetic dialogues to be generated automatically or semi-automatically.)
generating a first current prompt based on the current prompt template; (P0166, After templates (action retriever prompt template, response template) are created, slots to the templates are filled with diversified data such that the synthetic data covers varying complexities in different scenarios (e.g., 3 actions, 2 types of actions, and 4 slots to be filled). The slots can be used as the synthetic dataset for fine-tuning the routing LLM.)
invoking the pre-trained large language model based on the first current prompt to generate the generation content of the corresponding human-machine dialogue elements. (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM.)
Regarding claim 4 Hettige teaches claim 2 and further teaches:
wherein progressively generating the generation content of the corresponding human-machine dialogue elements based on the second prompt template, the pre-generated target generation content, and the pre-trained large language model according to the generation content dependency relationships comprises: determining, based on the generation content dependency relationships, a generation order of the generation content corresponding to the human-machine dialogue elements for the second prompt template; (P0009, The dialog script for the predefined scenario comprises an in-order dialog flow between a user and a digital assistant.; P0089, For example, in a dialog between a user and a digital assistant for an ordering-pizza task, the happy path would represent a conversation like,; P0090, User: I have an expense from a pizza restaurant.; P0091, Assistant: I need to know the amount and the date for this expense. Can you provide the amount first?; P0092, User: Sure, the amount was $200.)
sequentially executing second content generation operations from first to last according to the generation order, wherein the second content generation operations comprise:
obtaining, based on the generation content dependency relationships, pre-generated generation content on which the generation content of a current human-machine dialogue element depends, as the target generation content; (P0073, Context concerning the utterance are additionally appended to the output data and the utterance. The context is retrievable from the context and memory store and includes user session information, dialog state, conversation or contextual history, user information, or any combination thereof.)
formatting the second prompt template corresponding to the current human-machine dialogue element based on the target generation content, to generate a second current prompt; (P0166, After templates (action retriever prompt template, response template) are created, slots to the templates are filled with diversified data such that the synthetic data covers varying complexities in different scenarios (e.g., 3 actions, 2 types of actions, and 4 slots to be filled). The slots can be used as the synthetic dataset for fine-tuning the routing LLM.)
invoking the pre-trained large language model based on the second current prompt to generate the generation content of the current human-machine dialogue element. (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM.)
Regarding claim 5 Hettige teaches claim 1 and further teaches:
wherein the generation content dependency relationships are represented by placeholders set in the prompt templates. (P0193, The prompt template provides prompt placeholders associated with dialog script, such as the candidate actions, the context information, and/or an utterance.)
Regarding claim 6 Hettige teaches claim 5 and further teaches:
wherein the prompt templates represent, through placeholders corresponding to a first human-machine dialogue element, that input conditions on which the generation of the generation content of a second human-machine dialogue element depends include the generation content of the first human-machine dialogue element, wherein the second human-machine dialogue element is the human-machine dialogue element corresponding to the prompt template. (P0193, The prompt template provides prompt placeholders associated with dialog script, such as the candidate actions, the context information, and/or an utterance. A prompt placeholder associated with the context information may include at least a portion of an action plan (or an execution plan). For example, an action plan may include information about the action, agent, argument, date and time, and the like, and a prompt placeholder include information or seek for information where such information is missing from the action plan. The prompt placeholder may also include information related to the execution plan, which comprises an action including at least one argument slot having missing values (e.g., an execution plan with an action to acquire information from the user). The utterance may also be provided by the prompt template, which comprises information for filling in the missing values.)
Regarding claim 7 Hettige teaches claim 1 and further teaches:
wherein the prompt template corresponding to the user queries includes a rule prompt, the rule prompt being used to instruct the pre-trained large language model to generate user queries according to a first rule, wherein the first rule includes: each group of generated user queries has relevance and a logical progression, and incorporates at least N instances of context, where N is a natural number greater than 1. (P0100, Rule-based approaches can additionally or alternatively be used to synthesize data. These rules can be designed to reflect typical language usage within the domain being modeled. For example, fixed templates with placeholders can be used to create various dialogue scenarios. In some instances, conversations can be generated by defining how the conversations or sentences are structured using grammars or linguistic rules.; P0102, Simulated user interaction approaches involve generating dialogs mimicking how users might engage with a digital assistant. The approaches usually generate multi-turn, realistic conversations where users ask questions, make requests, or provide feedback. For example, simulating conversational flows allows a planner to script various user-initiated dialogues, such as booking a service, requesting recommendations, or asking for status updates. Similarly, task-oriented dialogue generation focuses on specific tasks like reserving a restaurant table or checking account balances. These simulated interactions create a broad set of conversational data, enabling the model to learn from diverse scenarios and improve its performance across different user interactions.)
Regarding claim 8 Hettige teaches claim 1 and further teaches:
wherein the prompt template corresponding to the response content includes an instruction prompt, the instruction prompt being used to instruct the pre-trained large language model to generate the response content to a final-round user query in dialogue context based on prior dialogue context. (P0073, The output data is appended to the utterance to construct an output prompt for input to the LLM. In some instances, context concerning the utterance are additionally appended to the output data and the utterance. The context is retrievable from the context and memory store and includes user session information, dialog state, conversation or contextual history, user information, or any combination thereof. The LLM generates responses based on the output prompt.; P0053, Digital assistant may end the conversation by generating a final response providing information to the user indicating that the pizza has been ordered.)
Regarding claim 9 Hettige teaches claim 1 and further teaches:
wherein the user queries have a logical progression, and the response content corresponds to the user queries, the multi-round dialogue data in the target scenario being generated based on the generation content corresponding to the user queries and the response content, comprising: for each group of user queries, constructing an ordered combination of multi-round question-answer pairs corresponding to each group of user queries based on the user queries and the corresponding response content, according to the logical progression of the user queries; (P0102, Simulated user interaction approaches involve generating dialogs mimicking how users might engage with a digital assistant. The approaches usually generate multi-turn, realistic conversations where users ask questions, make requests, or provide feedback. For example, simulating conversational flows allows a planner to script various user-initiated dialogues, such as booking a service, requesting recommendations, or asking for status updates. Similarly, task-oriented dialogue generation focuses on specific tasks like reserving a restaurant table or checking account balances.)
generating the multi-round dialogue data in the target scenario based on the ordered combination. (P0102, Generate multi-turn, realistic conversations.)
Regarding claim 10 Hettige teaches:
A method for training a model, comprising: obtaining multiple sets of multi-round dialogue data in a target scenario, wherein each set of the multi-round dialogue data comprises: (P0123, The learning algorithm is the overall method or procedure used to adjust the model parameters to fit the data.; P0117, Training is the initial phase of developing machine learning models where the model learns to make predictions, classifications, or decisions based on training data provided from the training and validation datasets.
P0129, Training an LLM using training data involves feeding the model vast amounts of text (or conversational data) to help it learn patterns, structures, and relationships in language.) multi-round user queries and response content arranged in a logically progressive sequence; (P0130, Fine-tuning the LLM is performed using training data including synthetic data in the training and validation dataset. For example, synthetic dialogs can be generated to simulate real-world conversations between users and a digital assistant. … These synthetic dialogs would encompass diverse user inputs, including requests for toppings, delivery time, or payment options. Fine-tuning the LLM on such synthetic data ensures the model understands the nuances of different domains or tasks (e.g., pizza ordering) and can respond accurately.; P0164, The synthetic data enhances the training data by adding secondary-happy-path data and non-sequitur data. For example, the synthetic data can be categorized into multi-action data (data involving in a scenario that relates to multiple actions, e.g., a conversation for finding restaurants and reserve a restaurant).)
constructing a dialogue content generation model based on a pre-trained large language model; (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM. This synthetic data can be created through a synthetic data generation pipeline, which is designed to populate a training dataset. The pipeline works by defining a series of dialogue scripts between a user and a digital assistant, complete with input/output templates and placeholders for specific slot values. These templates provide structured frameworks, allowing for a wide range of synthetic dialogues to be generated automatically or semi-automatically.) fine-tuning the dialogue content generation model based on the multiple sets of multi-round dialogue data until predicted loss value of the dialogue content generation model satisfies a preset convergence condition, wherein the predicted loss value is calculated based on single-round dialogue prediction loss of each set of multi-round dialogue data, and the single-round dialogue prediction loss is negative log-likelihood mean obtained by modeling predicted response content generated by the dialogue content generation model in response to the user queries in each round of dialogue data. (P0178, The model fine-tuning subsystem is also configured to define objective measurements to evaluate the performance of the model being fine-tuned. In some instances, the objective measurements include one or more loss functions that guide the learning process of the model, enabling optimization through methods such as gradient-based methods.; P0179, The loss functions can one or more of the followings: a cross-entropy loss, a mean squared error (MSE), a negative log-likelihood loss, a mean absolute error (MAE), a hinge loss, a Huber loss, and a combination of one or more of the loss functions. It should be understood that different loss functions or custom loss functions can be defined and used for fine-tuning the LLM module.; P0201, The fine-tuning process begins by generating batches of examples selected from the set of training examples and the set of synthesized training examples. An iterative training loop process in then performed for each batch of examples. The iterative training loop process first inputs examples from each batch into the pre-trained machine learning model. Losses are determined for sub-tasks of routing and slot-filling.; P0117, The predetermined optimization condition is achieved when a convergence criterion is met, such as when the change in the model parameters falls below a certain threshold between iterations.)
Regarding claim 11 Hettige teaches:
A method for processing a dialogue, comprising: (P0033, The method can include accessing a set of training examples, wherein each training example of the set of training examples includes a dialog script between a user and a digital assistant, generating a set of synthesized training examples using an iterative process that is performed for each of one or more predefined scenarios.)
in response to receiving a current-round user query, obtaining dialogue data from a specified number of previous rounds of dialogue; (P0073, The result of implementing the execution plan is output data (e.g., results of actions, data, information, etc.), which is transmitted to an output pipeline for generating end-user responses. For example, the output data from the assets (knowledge, API, dialog history, etc.) and relevant information from the context and memory store can be transmitted to the output pipeline.)
generating reply dialogue data to be responded to based on the current-round user query and the specified number of previous rounds of dialogue data, according to a logical progression; (0073, Context concerning the utterance are additionally appended to the output data and the utterance. The context is retrievable from the context and memory store and includes user session information, dialog state, conversation or contextual history, user information, or any combination thereof.)
invoking a pre-trained dialogue content generation model based on the reply dialogue data to obtain response content output by the pre-trained dialogue content generation model; (P0073, The output data is appended to the utterance to construct an output prompt for input to the LLM.; P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples.)
responding to the current-round user query based on the response content. (P0073, The LLM generates responses based on the output prompt.)
Regarding claim 12 Hettige teaches claim 11 and further teaches:
wherein the pre-trained dialogue content generation model is trained by: obtaining multiple sets of multi-round dialogue data in a target scenario, wherein each set of the multi-round dialogue data comprises: multi-round user queries and response content arranged in a logically progressive sequence; (P0102, Simulated user interaction approaches involve generating dialogs mimicking how users might engage with a digital assistant. The approaches usually generate multi-turn, realistic conversations where users ask questions, make requests, or provide feedback. For example, simulating conversational flows allows a planner to script various user-initiated dialogues, such as booking a service, requesting recommendations, or asking for status updates. Similarly, task-oriented dialogue generation focuses on specific tasks like reserving a restaurant table or checking account balances.)
constructing a dialogue content generation model based on a pre-trained large language model; (P0095, Pre-trained LLMs (e.g., an LLM other than the one being trained on synthetic data) can be leveraged to generate new data examples, such as secondary happy path examples, for the purpose of training or fine-tuning another LLM. This synthetic data can be created through a synthetic data generation pipeline, which is designed to populate a training dataset. The pipeline works by defining a series of dialogue scripts between a user and a digital assistant, complete with input/output templates and placeholders for specific slot values. These templates provide structured frameworks, allowing for a wide range of synthetic dialogues to be generated automatically or semi-automatically.) fine-tuning the dialogue content generation model based on the multiple sets of multi-round dialogue data until predicted loss value of the dialogue content generation model satisfies a preset convergence condition, wherein the predicted loss value is calculated based on single-round dialogue prediction loss of each set of multi-round dialogue data, and the single-round dialogue prediction loss is negative log-likelihood mean obtained by modeling predicted response content generated by the dialogue content generation model in response to the user queries in each round of dialogue data. (P0178, The model fine-tuning subsystem is also configured to define objective measurements to evaluate the performance of the model being fine-tuned. In some instances, the objective measurements include one or more loss functions that guide the learning process of the model, enabling optimization through methods such as gradient-based methods.; P0179, The loss functions can one or more of the followings: a cross-entropy loss, a mean squared error (MSE), a negative log-likelihood loss, a mean absolute error (MAE), a hinge loss, a Huber loss, and a combination of one or more of the loss functions. It should be understood that different loss functions or custom loss functions can be defined and used for fine-tuning the LLM module.; P0201, The fine-tuning process begins by generating batches of examples selected from the set of training examples and the set of synthesized training examples. An iterative training loop process in then performed for each batch of examples. The iterative training loop process first inputs examples from each batch into the pre-trained machine learning model. Losses are determined for sub-tasks of routing and slot-filling.; P0117, The predetermined optimization condition is achieved when a convergence criterion is met, such as when the change in the model parameters falls below a certain threshold between iterations.)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DANIEL WONSUK CHUNG whose telephone number is (571)272-1345. The examiner can normally be reached Monday - Friday (7am-4pm)[PT].
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, PIERRE-LOUIS DESIR can be reached at (571)272-7799. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/DANIEL W CHUNG/Examiner, Art Unit 2659
/PIERRE LOUIS DESIR/Supervisory Patent Examiner, Art Unit 2659
Prosecution Insights