DETAILED ACTION
Notice of AIA Status
The present application is being examined under the AIA the first inventor to file provisions.
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 07/09/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier.
Claim 17, recites limitations that use words like “means” (or “step”) or similar terms with functional language and do invoke 35 U.S.C. 112(f):
Claim 17; recites the limitation, “an allocation component configured to allocate resources” [Line 8].
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
After a careful analysis, as disclosed above, and a careful review of the specification the following limitations in claims 17:
“allocation component” (Fig. 3, #320. Paragraph [0043 and 0044]- allocation component 320 is an example of, or includes aspects of, the corresponding element described with reference to FIG. 7. According to some aspects, allocation component 320 is implemented as software stored in memory unit 310 and executable by processor unit 305, as firmware, as one or more hardware circuits, or as a combination thereof. (wherein the allocation component does have sufficient structure associated with it of software executed by a processor, firmware, a circuit, or any combination thereof.).)).
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
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 9 and its dependent claims 10-11 and 16 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more nor an integration of the judicial exceptions into a practical application. The limitations, under their broadest reasonable interpretation, cover mental process (concept performed in a human mind, including as observation, evaluation, judgment, opinion). The claimed invention simply perform training to determine a complexity of a user prompt. See analysis below for more details.
Regarding Independent Claim 9 and its dependent claims 10-16,
Step 1 Analysis: Claim 9 is directed to a method, which falls within one of the four statutory categories (process, machine, manufacture or composition of matter). Please see MPEP §2106.04.
Step 2A Prong 1 Analysis: Claim 9 recites, in part:
“generating, using a generative machine learning model, a synthetic output based on the training prompt;
and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt;
wherein the complexity value corresponds to an amount of resources for the generative machine learning model to achieve a target quality level based on the input prompt.”
The limitations as drafted, are processes that, under broadest reasonable interpretation, covers the performance of the limitation in the mind which falls within the “Mental Processes” grouping of abstract ideas. Please see MPEP §2106.04. The limitations of:
“generating, using a generative machine learning model, a synthetic output based on the training prompt” is a step a human mind can perform, under BRI, using pen and paper through a process of observation and evaluation such as getting a prompt from someone else or themselves and drawing what is described in the prompt.
“training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt” is a step a human mind can perform, under BRI, using pen and paper through a process of observation and evaluation such as determining how complex a prompt is.
“wherein the complexity value corresponds to an amount of resources for the generative machine learning model to achieve a target quality level based on the input prompt” is a step a human mind can perform, under BRI, using pen and paper through a process of observation and evaluation such as determining how much of a resource such as thought or effort to put into creating a prompt based on its assumed difficulty.
Notes: under MPEP 2106.04(a)(2)(III), mental process (thinking) “can be performed in the human mind, or by a human using a pen and paper" to be an abstract idea. CyberSource Corp. v. Retail Decisions, Inc., 654 F.3d 1366, 1372, 99 USPQ2d 1690, 1695 (Fed. Cir. 2011): "methods which can be performed mentally, or which are the equivalent of human mental work, are unpatentable abstract ideas the ‘basic tools of scientific and technological work’ that are open to all." (citing Gottschalk v. Benson, 409 U.S. 63, 175 USPQ 673 [1972]). See also Mayo Collaborative Servs. v. Prometheus Labs. Inc., 566 U.S. 66, 71, 101 USPQ2d 1961, 1965 ("mental processes and abstract intellectual concepts are not patentable, as they are the basic tools of scientific and technological work’" (quoting Benson, 409 U.S. at 67, 175 USPQ at 675).
The courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide rule) to perform the claim limitation. See, e.g., Benson, 409 U.S. at 67, 65, 175 USPQ at 674-75, 674; Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1139, 120 USPQ2d 1473, 1474 (Fed. Cir. 2016).
Nor do the courts distinguish between claims that recite mental processes performed by humans and claims that recite mental processes performed on a computer, generic circuit or device, or the likes. See " Versata Dev. Group v. SAP Am., Inc., 793 F.3d 1306, 1335, 115 USPQ2d 1681, 1702 (Fed. Cir. 2015). See also Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1318, 120 USPQ2d 1353, 1360 (Fed. Cir. 2016) (‘with the exception of generic computer-implemented steps, there is nothing in the claims themselves that foreclose them from being performed by a human, mentally or with pen and paper.’’).
Because both product/device and process claims may recite a "mental process", the phrase "mental processes" should be understood as referring to the type of abstract idea, and not to the statutory category of the claim. The courts have identified numerous product claims as reciting mental process-type abstract ideas, for instance the product claims to computer systems and computer-readable media in Versata Dev. Group. v. SAP Am., Inc., 793 F.3d 1306, 115 USPQ2d 1681 (Fed. Cir. 2015).
Accordingly, the claim recites an abstract idea.
Step 2A Prong 2 Analysis: This judicial exception is not integrated into a practical application. particular, the claim recites the following additional element(s) –
“obtaining a training set including a training prompt”
The additional elements “obtaining a training set including a training prompt” include steps of insignificant extra-solution/post-solution activities of data gathering, data generating, data transmitting, etc.
Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claim as a whole is directed to an abstract idea. Please see MPEP §2106.04. (d).III.C.
Step 2B Analysis: there are no additional elements, such as for these additional elements as indicated above, that amount to significantly more than the judicial exception. Please see MPEP §2106.05. The claim is directed to an abstract idea. Please see MPEP §2106.05
For all of the foregoing reasons, claim 9 does not comply with the requirements of 35 USC 101.
Accordingly, the dependent claims 10-11 and 16 do not provide elements that overcome the deficiencies of the independent claim 9.
Moreover, claim 10 recites, in part,
“determining a quality value of the synthetic output, wherein the classifier network is trained based on the quality value.”
Which is a mental process activity abstract idea of observation and evaluation, judgement, merely performing a determination of the quality of the output for a given input and learning based on its quality.
Moreover, claim 11 recites, in part,
“wherein determining the quality value comprises: comparing the synthetic output to a ground-truth media asset.”
Which is a mental process activity abstract idea of observation and evaluation, judgement, merely performing a comparison between the output and an expected value.
Moreover, claim 16 recites, in part,
“generating, using the classifier network, a predicted complexity value based on the training prompt”
Which is a mental process activity abstract idea of observation and evaluation, judgement, merely performing generation of an expected complexity value by looking at a training prompt.
“comparing the predicted complexity value to a ground-truth complexity value for the training prompt.”
Which is a mental process activity abstract idea of observation and evaluation, judgement, merely performing a comparison between the output and an expected value for how complex an input is.
Accordingly, the dependent claims 10-11 and 16 are not patent eligible under 101.
Moreover, claim 12 recites, in part,
“generating a plurality of synthetic outputs based on the training prompt using a plurality of different resource allocations, respectively”
“and selecting a target resource allocation from among the plurality of different resource allocations based on the plurality of synthetic outputs, wherein the classifier network is trained based on the target resource allocation.”
This would overcome the 101 because of “generating a plurality of synthetic outputs based on the training prompt using a plurality of different resource allocations, respectively; and selecting a target resource allocation from among the plurality of different resource allocations based on the plurality of synthetic outputs”
Accordingly, the dependent claim 12 and its dependent claims 13-15 are patent eligible under 101.
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.
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
Claims 1, 4-5, 8-10, 12, 14-15, 17, and 19, are rejected under 35 U.S.C. 102(a)(1)/(a)(2) as being anticipated by Cook et al. (US 20250272313 A1) hereafter referenced as Cook.
Regarding claim 1, Cook teaches a method for generative machine learning (Fig. 1, paragraph [0014]- Cook discloses examples described in this disclosure relate to systems and methods for selecting an AI model, such as a large language model (LLM), multimodal model, or other type of generative AI model, based on a query complexity.),
comprising: obtaining an input prompt (Fig. 2, Paragraph [0034]- Cook discloses at operation 202, an input query is received.);
generating, using a classifier network, a complexity value of the input prompt (Fig. 3, Paragraph [0037]- Cook discloses the response model selector 306 includes a response classifier 308 that is configured (e.g., trained) to classify the input query as being associated with one of a predetermined set of response complexity scores (e.g., to label the input query with an appropriate response complexity score).),
wherein the complexity value corresponds to an amount of resources for a generative machine learning model to achieve a target quality level based on the input prompt (Fig. 3, Paragraph [0038]- Cook discloses if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response.);
allocating resources of the generative machine learning model based on the complexity value (Fig. 3, Paragraph [0039]- Cook discloses after the response classifier 308 determines a response complexity score for the input query, the response model selector 104 selects the first AI model 314 or the second AI model 316 for generating a response to the input query by comparing the response complexity score to a threshold score.);
and generating, using the generative machine learning model, a synthetic output based on the input prompt using the allocated resources (Fig. 3, Paragraph [0043]- Cook discloses the response model selector 306 selects an AI model for generating a response to a query based on a combination of criteria that include the response complexity score of the query (e.g., compared to a threshold score), the utilization of the AI model(s), and/or the AI model selected for previous queries in the conversation.),
wherein the synthetic output has the target quality level (Fig. 3, Paragraph [0038]- Cook discloses if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response.).
Regarding claim 4, Cook teaches the method of claim 1, Cook further teaches wherein allocating the resources comprises: determining a size of the generative machine learning model (Fig. 3, Paragraph [0036]- Cook discloses the response model selector 306 is configured to analyze an input query received via the chat interface 302 (along with any additional relevant context, such as prior input queries in the same conversation and/or prior responses generated in response to prior input queries) to determine a complexity score associated with the input query and select, based on the complexity score, a first AI model 314 or a second AI model 316 for generating a response to the input query. In the current example, the first AI model 314 and second AI model 316 are LLM-based models such as described with reference to the AI models 110,112 of FIG. 1, with the first AI model 314 being a higher-complexity model than the second AI model 316 (e.g., the first AI model 314 has more parameters than the second AI model 316 and/or other differences in performance characteristics) (wherein the number of parameters is seen as the size of the model).).
Regarding claim 5, Cook teaches the method of claim 1, Cook further teaches wherein allocating the resources comprises: selecting the generative machine learning model from among a plurality of candidate machine learning models (Fig. 3, Paragraph [0036]- Cook discloses the response model selector 306 is configured to analyze an input query received via the chat interface 302 (along with any additional relevant context, such as prior input queries in the same conversation and/or prior responses generated in response to prior input queries) to determine a complexity score associated with the input query and select, based on the complexity score, a first AI model 314 or a second AI model 316 for generating a response to the input query.).
Regarding claim 8, Cook teaches the method of claim 1, Cook further teaches wherein: the classifier network is trained by determining a quality of an output of the generative machine learning model (Fig. 3, Paragraph [0038]- Cook discloses such training queries may be extracted from logs of prior queries that have been submitted via the chat interface. The evaluation AI model (which may be an LLM or other generative AI model) is configured to evaluate (e.g., determine) the absolute or relative quality of the responses based on various quality metrics to enable comparisons of response quality across the Al models.).
Regarding claim 9, Cook teaches a method for training a machine learning model, comprising (Fig. 1, paragraph [0026]- Cook discloses the language model is generally trained using supervised learning based on large amounts of annotated text data.):
obtaining a training set including a training prompt (Fig. 3 Paragraph [0038]- Cook discloses the training dataset may be generated by providing each query of the set of training queries (which may be prior queries that have been collected for this purpose) to multiple AI models of differing complexity (including, in some examples, the first AI model 314 and/or the second AI model 316) and providing the queries and resulting responses to an evaluation AI model (not shown).);
generating, using a generative machine learning model, a synthetic output based on the training prompt (Fig. 3 Paragraph [0038]- Cook discloses the training dataset may be generated by providing each query of the set of training queries (which may be prior queries that have been collected for this purpose) to multiple AI models of differing complexity (including, in some examples, the first AI model 314 and/or the second AI model 316) and providing the queries and resulting responses to an evaluation AI model (not shown).);
and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt (Fig. 3, paragraph [0038]- Cook discloses in this manner, a set of query and response complexity score pairs can be generated and used to train the response classifier 308 such that the response classifier 308 is able to determine a response complexity score for a new input query (e.g., an input query that is not included in the training data).),
wherein the complexity value corresponds to an amount of resources for the generative machine learning model to achieve a target quality level based on the input prompt (Fig. 3, Paragraph [0038]- Cook discloses if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response.).
Regarding claim 10, Cook teaches the method of claim 9, Cook further teaches further comprising: determining a quality value of the synthetic output (Fig. 3, Paragraph [0038]- Cook discloses the evaluation AI model determines, based on the quality metrics, a quality score for each response.),
wherein the classifier network is trained based on the quality value (Fig. 3, Paragraph [0038]- Cook discloses the quality scores, in turn, are used to assign a response complexity score to the query. For example, if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response. Conversely, if the lower-complexity AI model generates a response that is determined to have a similar quality score as that of the response generated using the higher-complexity AI model, the input query is assigned a relatively low response complexity score, indicating that the query may not need a higher-complexity AI model to generate an acceptable response. In this manner, a set of query and response complexity score pairs can be generated and used to train the response classifier 308 such that the response classifier 308 is able to determine a response complexity score for a new input query (e.g., an input query that is not included in the training data).).
Regarding claim 12, Cook teaches the method of claim 9, Cook further teaches further comprising: generating a plurality of synthetic outputs based on the training prompt using a plurality of different resource allocations, respectively (Fig. 3, Paragraph [0038]- Cook discloses the training dataset may be generated by providing each query of the set of training queries (which may be prior queries that have been collected for this purpose) to multiple AI models of differing complexity (including, in some examples, the first AI model 314 and/or the second AI model 316) and providing the queries and resulting responses to an evaluation AI model (not shown).);
and selecting a target resource allocation from among the plurality of different resource allocations based on the plurality of synthetic outputs (Fig. 3, Paragraph [0038]- Cook discloses the quality scores, in turn, are used to assign a response complexity score to the query. For example, if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response. Conversely, if the lower-complexity AI model generates a response that is determined to have a similar quality score as that of the response generated using the higher-complexity AI model, the input query is assigned a relatively low response complexity score, indicating that the query may not need a higher-complexity AI model to generate an acceptable response.),
wherein the classifier network is trained based on the target resource allocation (Fig. 3, Paragraph [0038]- Cook discloses in this manner, a set of query and response complexity score pairs can be generated and used to train the response classifier 308 such that the response classifier 308 is able to determine a response complexity score for a new input query (e.g., an input query that is not included in the training data).).
Regarding claim 14, Cook teaches the method of claim 12, Cook further teaches wherein selecting the target resource allocation comprises: generating the plurality of synthetic outputs until a quality condition is satisfied (Fig. 3, paragraph [0038]- Cook discloses the training dataset may be generated by providing each query of the set of training queries (which may be prior queries that have been collected for this purpose) to multiple AI models of differing complexity (including, in some examples, the first AI model 314 and/or the second AI model 316) and providing the queries and resulting responses to an evaluation AI model (not shown) (wherein the quality condition being satisfied is the comparison between the multiple responses).),
wherein the target resource allocation is selected based on resources allocated to the generative machine learning model when the quality condition is satisfied (Fig. 3, paragraph [0038]- Cook discloses the evaluation AI model determines, based on the quality metrics, a quality score for each response. The quality scores, in turn, are used to assign a response complexity score to the query. For example, if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response. Conversely, if the lower-complexity AI model generates a response that is determined to have a similar quality score as that of the response generated using the higher-complexity AI model, the input query is assigned a relatively low response complexity score, indicating that the query may not need a higher-complexity AI model to generate an acceptable response (wherein the target resource allocation selected the complexity score assigned to the model)).
Regarding claim 15, Cook teaches the method of claim 12, Cook further teaches further comprising: determining a training complexity value based on the target resource allocation (Fig. 3, Paragraph [0036]- Cook discloses the evaluation AI model determines, based on the quality metrics, a quality score for each response. The quality scores, in turn, are used to assign a response complexity score to the query.).
Regarding claim 17, Cook teaches a system for generative machine learning (Fig. 1, paragraph [0014]- Cook discloses examples described in this disclosure relate to systems and methods for selecting an AI model, such as a large language model (LLM), multimodal model, or other type of generative AI model, based on a query complexity.),
comprising: at least one memory (Fig. 7, Paragraph [0084]- Cook discloses the computing device components described below may be suitable for one or more of the components of the systems described above. In a basic configuration, the computing device 700 includes at least one processing unit 702 and a system memory 704.);
at least one processor executing instructions stored in the at least one memory (Fig. 7, paragraph [0086]- Cook discloses a number of program modules and data files may be stored in the system memory 704. While executing on the processing unit 702, the program modules 706 may perform processes including one or more of the stages of the methods 200, 500, and/or 600, illustrated in FIGS. 2, 5, 6A, and 6B.);
a classifier network comprising classification parameters stored in the at least one memory (Fig. 1, Paragraph [0099]- Cook discloses the functionality associated with some examples described in this disclosure can also include instructions stored in a non-transitory media. The term “non-transitory media” as used herein refers to any media storing data and/or instructions that cause a machine to operate in a specific manner.),
the classifier network trained to generate a complexity value of an input prompt (Fig. 3, Paragraph [0037]- Cook discloses the response model selector 306 includes a response classifier 308 that is configured (e.g., trained) to classify the input query as being associated with one of a predetermined set of response complexity scores (e.g., to label the input query with an appropriate response complexity score).),
wherein the complexity value corresponds to an amount of resources to achieve a target quality level based on the input prompt (Fig. 3, Paragraph [0038]- Cook discloses if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response.);
an allocation component configured to allocate resources based on the complexity value (Fig. 3, Paragraph [0039]- Cook discloses after the response classifier 308 determines a response complexity score for the input query, the response model selector 104 selects the first AI model 314 or the second AI model 316 for generating a response to the input query by comparing the response complexity score to a threshold score.);
and a generative machine learning model comprising generative parameters stored in the at least one memory (Fig. 1, Paragraph [0099]- Cook discloses the functionality associated with some examples described in this disclosure can also include instructions stored in a non-transitory media. The term “non-transitory media” as used herein refers to any media storing data and/or instructions that cause a machine to operate in a specific manner.),
the generative machine learning model trained to generate a synthetic output based on the input prompt using the allocated resources (Fig. 3, Paragraph [0043]- Cook discloses the response model selector 306 selects an AI model for generating a response to a query based on a combination of criteria that include the response complexity score of the query (e.g., compared to a threshold score), the utilization of the AI model(s), and/or the AI model selected for previous queries in the conversation.),
wherein the synthetic output has the target quality level (Fig. 3, Paragraph [0038]- Cook discloses if a higher-complexity AI model generates a response that is determined to have a significantly higher quality score than that of the response generated using a lower-complexity AI model, the input query is assigned a relatively high response complexity score, indicating that the query is likely to need a higher-complexity Al model to generate a high-quality response.).
Regarding claim 19, Cook teaches the system of claim 17, Cook further teaches wherein: the generative machine learning model comprises a configurable number of parameters (Fig. 3, Paragraph [0036]- Cook discloses the response model selector 306 is configured to analyze an input query received via the chat interface 302 (along with any additional relevant context, such as prior input queries in the same conversation and/or prior responses generated in response to prior input queries) to determine a complexity score associated with the input query and select, based on the complexity score, a first AI model 314 or a second AI model 316 for generating a response to the input query. In the current example, the first AI model 314 and second AI model 316 are LLM-based models such as described with reference to the AI models 110,112 of FIG. 1, with the first AI model 314 being a higher-complexity model than the second AI model 316 (e.g., the first AI model 314 has more parameters than the second AI model 316 and/or other differences in performance characteristics),
wherein the allocated resources indicates a value for the configurable number of parameters (Fig. 3, Paragraph [0036]- Cook discloses the response model selector 306 is configured to analyze an input query received via the chat interface 302 (along with any additional relevant context, such as prior input queries in the same conversation and/or prior responses generated in response to prior input queries) to determine a complexity score associated with the input query and select, based on the complexity score, a first AI model 314 or a second AI model 316 for generating a response to the input query. In the current example, the first AI model 314 and second AI model 316 are LLM-based models such as described with reference to the AI models 110,112 of FIG. 1, with the first AI model 314 being a higher-complexity model than the second AI model 316 (e.g., the first AI model 314 has more parameters than the second AI model 316 and/or other differences in performance characteristics).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 2-3 and 18 are rejected under 35 U.S.C 103 as being unpatentable Cook et al. (US 20250272313 A1) hereafter referenced as Cook in view of Westcott et al. (US 20250133238 A1) hereafter referenced as Westcott.
Regarding claim 2, Cook teaches the method of claim 1, Cook fails to explicitly teach wherein allocating the resources comprises: determining a diffusion time step based on the complexity value.
However, Westcott explicitly teaches wherein allocating the resources comprises: determining a diffusion time step based on the complexity value (Fig. 1, Paragraph [0113]- Westcott discloses once trained, the number of diffusion steps between frames may vary. The number of diffusion steps could vary based on the raw framerate, or it could dynamically change based on changes in the image.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for generative machine learning, comprising: obtaining an input prompt; generating, using a classifier network, a complexity value of the input prompt with the teachings of Westcott wherein allocating the resources comprises: determining a diffusion time step based on the complexity value.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein allocating the resources comprises: determining a diffusion time step based on the complexity value.
The motivation behind the modification would have been to allow for reduced the scaling cost of computation, since both Cook and Westcott are both systems that dynamically adjust resources based on complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Westcott’s system provides a way to reduce the scaling cost of computation while maintaining the same speed. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Westcott et al. (US 20250133238 A1) Paragraph [0113].
Regarding claim 3, Cook in view of Westcott teaches the method of claim 2, Cook fails to explicitly teach wherein generating the synthetic output comprises: performing a diffusion process based on a noise input, the input prompt, and the diffusion time step.
However, Westcott explicitly teaches wherein generating the synthetic output comprises: performing a diffusion process based on a noise input (Fig. 1, Paragraph [0109]- Westcott modern denoising diffusion models typically slowly add noise to a target image with a well-defined distribution (e.g., Gaussian) to transform it from a structured image to noise in the forward process, allowing a ML model to learn the information needed to reconstruct the image from noise in the reverse process.),
the input prompt (Fig. 1, Paragraph [0054]- Westcott discloses the diffusion model 124 is conditionally trained using image frames 115 captured prior to or during the training phase and conditioning data 117 derived from the training image frames by a conditioning data extraction module 116 (wherein the input prompt are the frames input).),
and the diffusion time step (Fig. 1, Paragraph [0113]- Westcott discloses once trained, the number of diffusion steps between frames may vary. The number of diffusion steps could vary based on the raw framerate, or it could dynamically change based on changes in the image.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook in view of Westcott of having a method for generative machine learning, comprising: obtaining an input prompt; generating, using a classifier network, a complexity value of the input prompt with the teachings of Westcott wherein generating the synthetic output comprises: performing a diffusion process based on a noise input, the input prompt, and the diffusion time step.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein generating the synthetic output comprises: performing a diffusion process based on a noise input, the input prompt, and the diffusion time step.
The motivation behind the modification would have been to allow for reduced the scaling cost of computation, since both Cook and Westcott are both systems that dynamically adjust resources based on complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Westcott’s system provides a way to reduce the scaling cost of computation while maintaining the same speed. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Westcott et al. (US 20250133238 A1) Paragraph [0113].
Regarding claim 18, Cook teaches the system of claim 17, Cook further teaches wherein: the generative machine learning model comprises an image generation model (Fig. 1, Paragraph [0024]- Cook discloses the input query (and optionally, additional context such as previous queries and/or responses) are initially processed during a reasoning stage 106, in which a first AI model 110 analyzes the input query to identify which tools 114 (if any) may be relevant to the input query; e.g., which tools should be used to gather information that may be needed to generate a response to the input query. Such tools may include, for example, web search tools, image creation tools, image understanding tools, advertisement generation tools, third-party plugins, restaurant reservation tools, code generation tools, map tools, or other types of tools.),
Cook fails to explicitly teach and the allocated resources comprise a number of image generation steps.
However, Westcott explicitly teaches and the allocated resources comprise a number of image generation steps (Fig. 1, Paragraph [0113]- Westcott discloses once trained, the number of diffusion steps between frames may vary. The number of diffusion steps could vary based on the raw framerate, or it could dynamically change based on changes in the image (wherein the number of diffusion steps is seen as the number of image generation steps).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a system for generative machine learning, comprising: at least one memory; at least one processor executing instructions stored in the at least one memory; a classifier network comprising classification parameters stored in the at least one memory, the classifier network trained to generate a complexity value of an input prompt with the teachings of Westcott the allocated resources comprise a number of image generation steps.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein the allocated resources comprise a number of image generation steps.
The motivation behind the modification would have been to allow for reduced the scaling cost of computation, since both Cook and Westcott are both systems that dynamically adjust resources based on complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Westcott’s system provides a way to reduce the scaling cost of computation while maintaining the same speed. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Westcott et al. (US 20250133238 A1) Paragraph [0113].
Claims 6 and 20 are rejected under 35 U.S.C 103 as being unpatentable Cook et al. (US 20250272313 A1) hereafter referenced as Cook in view of Nyamwange et al. (US 20250037005 A1) hereafter referenced as Nyamwange.
Regarding claim 6, Cook teaches the method of claim 1, Cook fails to explicitly teach wherein allocating the resources comprises: selecting a processor for generating the synthetic output
However, Nyamwange explicitly teaches wherein allocating the resources comprises: selecting a processor for generating the synthetic output (Fig. 2, paragraph [0060]- Nyamwange discloses allocating the subset of computational resources to the ML model. In this effort, the process flow may include determining a group of cores from the plurality of processing units. Having determined the group of cores, the process flow may include allocating the group of cores to the ML model.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for generative machine learning, comprising: obtaining an input prompt; generating, using a classifier network, a complexity value of the input prompt with the teachings of Nyamwange wherein allocating the resources comprises: selecting a processor for generating the synthetic output.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein allocating the resources comprises: selecting a processor for generating the synthetic output
The motivation behind the modification would have been to allow improved performance and efficiency in a distributed computing environment, since both Cook and Nyamwange are both systems that dynamically adjust resources used by a Machine learning model. Wherein Cook’s system wherein minimized the resource consumption of the system, while Nyamwange’s system provides a way to improve distribution performance of the system allowing for load balancing reducing computational resource usage and redistribution of computational load. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Nyamwange et al. (US 20250037005 A1) Paragraph [0037 and 0068].
Regarding claim 20, Cook teaches the system of claim 17, Cook fails to explicitly teach the system further comprising: a plurality of processors, wherein the allocated resources comprises one or more of the plurality of processors.
However, Nyamwange explicitly teaches the system further comprising: a plurality of processors, wherein the allocated resources comprises one or more of the plurality of processors (Fig. 2, paragraph [0060]- Nyamwange discloses allocating the subset of computational resources to the ML model. In this effort, the process flow may include determining a group of cores from the plurality of processing units. Having determined the group of cores, the process flow may include allocating the group of cores to the ML model.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a system for generative machine learning, comprising: at least one memory; at least one processor executing instructions stored in the at least one memory; a classifier network comprising classification parameters stored in the at least one memory, the classifier network trained to generate a complexity value of an input prompt with the teachings of Nyamwange the system further comprising: a plurality of processors, wherein the allocated resources comprises one or more of the plurality of processors.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein the system further comprising: a plurality of processors, wherein the allocated resources comprises one or more of the plurality of processors.
The motivation behind the modification would have been to allow improved performance and efficiency in a distributed computing environment, since both Cook and Nyamwange are both systems that dynamically adjust resources used by a Machine learning model. Wherein Cook’s system wherein minimized the resource consumption of the system, while Nyamwange’s system provides a way to improve distribution performance of the system allowing for load balancing reducing computational resource usage and redistribution of computational load. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Nyamwange et al. (US 20250037005 A1) Paragraph [0037 and 0068].
Claim 7 is rejected under 35 U.S.C 103 as being unpatentable Cook et al. (US 20250272313 A1) hereafter referenced as Cook in view of Fedyk et al. (US 20260162327 A1) hereafter referenced as Fedyk.
Regarding claim 7, Cook teaches the method of claim 1, Cook further teaches wherein: the generative machine learning model comprises an image generation model (Fig. 1, Paragraph [0024]- Cook discloses the input query (and optionally, additional context such as previous queries and/or responses) are initially processed during a reasoning stage 106, in which a first AI model 110 analyzes the input query to identify which tools 114 (if any) may be relevant to the input query; e.g., which tools should be used to gather information that may be needed to generate a response to the input query. Such tools may include, for example, web search tools, image creation tools, image understanding tools, advertisement generation tools, third-party plugins, restaurant reservation tools, code generation tools, map tools, or other types of tools.),
Cook is silent to explicitly teach and the synthetic output comprises an image that depicts an element described by the input prompt.
However, Fedyk explicitly teaches and the synthetic output comprises an image that depicts an element described by the input prompt (Fig. 7, paragraph [0102]- Fedyk discloses the image-generation machine-learning model outputs a generated image that is responsive to the text prompt, where the generated image includes a depiction of the entity in the transcribed text.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for generative machine learning, comprising: obtaining an input prompt; generating, using a classifier network, a complexity value of the input prompt with the teachings of Fedyk the synthetic output comprises an image that depicts an element described by the input prompt.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein the synthetic output comprises an image that depicts an element described by the input prompt.
The motivation behind the modification would have been to allow for improved image generation, since both Cook and Fedyk are both systems that generate images using generative machine learning. Wherein Cook’s system wherein minimized the resource consumption of the system, while Fedyk’s system provides a way to improve efficiency of image generation. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Fedyk et al. (US 20260162327 A1) Paragraph [0024].
Claims 11 and 16 are rejected under 35 U.S.C 103 as being unpatentable Cook et al. (US 20250272313 A1) hereafter referenced as Cook in view of Bahirwani et al. (US 20250232141 A1) hereafter referenced as Bahirwani.
Regarding claim 11, Cook teaches the method of claim 10, Cook fails to explicitly teach wherein determining the quality value comprises: comparing the synthetic output to a ground-truth media asset.
However, Bahirwani explicitly teaches wherein determining the quality value comprises: comparing the synthetic output to a ground-truth media asset (Fig. 1, Paragraph [0046]- Bahirwani discloses when errors occur between the generated complexity as compared to a ground truth complexity of the training data, a type and/or degree of the error may be used by the training engine 114 in a subsequent training iteration to adjust weights or other parameters of the complexity model 122.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for training a machine learning model, comprising: obtaining a training set including a training prompt; generating, using a generative machine learning model, a synthetic output based on the training prompt; and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt with the teachings of Bahirwani wherein determining the quality value comprises: comparing the synthetic output to a ground-truth media asset.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein determining the quality value comprises: comparing the synthetic output to a ground-truth media asset.
The motivation behind the modification would have been to allow improved the accuracy of the trained models, since both Cook and Bahirwani are both systems that train machine learning models to analyze prompts for complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Bahirwani’s system provides a way to improve the accuracy of the trained models. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Bahirwani et al. (US 20250232141 A1) Paragraph [0053].
Regarding claim 16, Cook teaches the method of claim 9, Cook further teaches wherein training the classifier network comprises: generating, using the classifier network, a predicted complexity value based on the training prompt (Fig. 3, Paragraph [0038]- Cook discloses the evaluation AI model determines, based on the quality metrics, a quality score for each response. The quality scores, in turn, are used to assign a response complexity score to the query.);
Cook fails to explicitly teach and comparing the predicted complexity value to a ground-truth complexity value for the training prompt.
However, Bahirwani explicitly teaches and comparing the predicted complexity value to a ground-truth complexity value for the training prompt (Fig. 1, Paragraph [0046]- Bahirwani discloses when errors occur between the generated complexity as compared to a ground truth complexity of the training data, a type and/or degree of the error may be used by the training engine 114 in a subsequent training iteration to adjust weights or other parameters of the complexity model 122.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for training a machine learning model, comprising: obtaining a training set including a training prompt; generating, using a generative machine learning model, a synthetic output based on the training prompt; and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt with the teachings of Bahirwani comparing the predicted complexity value to a ground-truth complexity value for the training prompt.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein comparing the predicted complexity value to a ground-truth complexity value for the training prompt.
The motivation behind the modification would have been to allow improved the accuracy of the trained models, since both Cook and Bahirwani are both systems that train machine learning models to analyze prompts for complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Bahirwani’s system provides a way to improve the accuracy of the trained models. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Bahirwani et al. (US 20250232141 A1) Paragraph [0053].
Claim 13 is rejected under 35 U.S.C 103 as being unpatentable Cook et al. (US 20250272313 A1) hereafter referenced as Cook in view of Westcott et al. (US 20250133238 A1) hereafter referenced as Westcott and Nyamwange et al. (US 20250037005 A1) hereafter referenced as Nyamwange.
Regarding claim 13, Cook teaches the method of claim 12, Cook further teaches a network size, or any combination thereof (Fig. 3, Paragraph [0036]- Cook discloses the response model selector 306 is configured to analyze an input query received via the chat interface 302 (along with any additional relevant context, such as prior input queries in the same conversation and/or prior responses generated in response to prior input queries) to determine a complexity score associated with the input query and select, based on the complexity score, a first AI model 314 or a second AI model 316 for generating a response to the input query. In the current example, the first AI model 314 and second AI model 316 are LLM-based models such as described with reference to the AI models 110,112 of FIG. 1, with the first AI model 314 being a higher-complexity model than the second AI model 316 (e.g., the first AI model 314 has more parameters than the second AI model 316 and/or other differences in performance characteristics) (wherein the number of parameters is seen as the size of the model).).
Cook fails to explicitly teach wherein: the target resource allocation comprises a diffusion time step.
However, Westcott explicitly teaches wherein: the target resource allocation comprises a diffusion time step (Fig. 1, Paragraph [0113]- Westcott discloses once trained, the number of diffusion steps between frames may vary. The number of diffusion steps could vary based on the raw framerate, or it could dynamically change based on changes in the image.),
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for training a machine learning model, comprising: obtaining a training set including a training prompt; generating, using a generative machine learning model, a synthetic output based on the training prompt; and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt with the teachings of Westcott wherein: the target resource allocation comprises a diffusion time step.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein: the target resource allocation comprises a diffusion time step.
The motivation behind the modification would have been to allow for reduced the scaling cost of computation, since both Cook and Westcott are both systems that dynamically adjust resources based on complexity. Wherein Cook’s system wherein minimized the resource consumption of the system, while Westcott’s system provides a way to reduce the scaling cost of computation while maintaining the same speed. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Westcott et al. (US 20250133238 A1) Paragraph [0113].
Cook in view of Westcott fails to explicitly teach a processor.
However, Nyamwange explicitly teaches a processor (Fig. 2, paragraph [0060]- Nyamwange discloses allocating the subset of computational resources to the ML model. In this effort, the process flow may include determining a group of cores from the plurality of processing units. Having determined the group of cores, the process flow may include allocating the group of cores to the ML model.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Cook of having a method for training a machine learning model, comprising: obtaining a training set including a training prompt; generating, using a generative machine learning model, a synthetic output based on the training prompt; and training, using the training set and the synthetic output, a classifier network to generate a complexity value of an input prompt with the teachings of Nyamwange a processor.
Wherein having Cook’s system for allocating resources based on complexity of a user prompt wherein a processor.
The motivation behind the modification would have been to allow improved performance and efficiency in a distributed computing environment, since both Cook and Nyamwange are both systems that dynamically adjust resources used by a Machine learning model. Wherein Cook’s system wherein minimized the resource consumption of the system, while Nyamwange’s system provides a way to improve distribution performance of the system allowing for load balancing reducing computational resource usage and redistribution of computational load. Please see Cook et al. (US 20250272313 A1), Paragraph [0015] and Nyamwange et al. (US 20250037005 A1) Paragraph [0037 and 0068].
Conclusion
Listed below are the prior arts made of record and not relied upon but are considered
pertinent to applicant`s disclosure.
Leber et al. (US 9881088 B1)- Natural language solution generating devices and methods are provided herein. Exemplary devices may execute logic via one or more processors, which are programmed to receive a complex query in natural language format, the complex query including a real-world problem that requires interrogation of a plurality of information sources in order to ascertain a response to the problem, evaluate the complex query to determine query segments, which are each included with at least one domain, wherein a domain corresponds to an information source, query the information sources to obtain responses for the query segments, and generate a natural language solution using the responses....................Please see Fig. 1. Abstract.
Stefani et al. (US 20230196199 A1)- Querying databases may be performed with references to machine learning models. A database query may be received that references a machine learning model and database. In response to the query, the machine learning model may provide information which may be returned as part of a result of the query or may be used to generate a result of the query. The machine learning model may be generated in response to a request to generate a machine learning model that includes a database query that identifies the data upon which a machine learning technique may be applied to generate the machine learning model......................Please see Fig. 1. Abstract.
Sureka et al. (US 12260657 B1)- Presented herein are systems and methods for the employment of machine learning models for image processing. A method may include a capture of a video feed including image data of a document at a client device. The client device can provide the video feed to another computing device. The method can include, by the client device or the other computing device object recognition for recognizing a type of document and capturing an image exceeding a quality threshold of the document amongst the frames within the video feed. The method may further include the execution of other image processing operations on the image data to improve the quality of the image or features extracted therefrom. The method may further include anti-fraud detection or scoring operations to determine an amount of risk associated with the image data......................Please see Fig. 1. Abstract.
Relic et al. (US 20250157087 A1)- In some embodiments, a method receives a quantized latent representation of an image in a latent space. The image is encoded into a representation in the latent space and quantized to generate the quantized latent representation. A time step parameter is received that is generated based on the representation. The method performs an inverse quantization process to generate a reconstructed representation. A diffusion model performs a denoising process for a number of iterations based on the time step parameter to remove noise from the reconstructed representation to generate a denoised reconstructed representation. The denoised reconstructed representation is decoded into a reconstructed image.......................Please see Fig. 1. Abstract.
Maker et al. (US 20250138888 A1)- In one aspect, disclosed in a method for use in connection with a local area network (LAN) system comprising a group of multiple devices that includes a first device and a separate set of devices. The method includes: the first device determining that a machine learning (ML)-based task is to be performed; the first device broadcasting to the separate set of devices, a ML-based task request for the ML-based task, (i) wherein the separate set of devices are configured to perform an arbitration process to select a second device, and (ii) wherein the second device is configured to perform the ML-based task in accordance with the ML-based task request, thereby generating ML-based task output, and to transmit the generated ML-based task output to the first device; and the first device receiving the generated output from the second device and using the received output to facilitate performing one or more operations.......................Please see Fig. 1. Abstract.
Baracaldo Angel et al. (US 20170090975 A1)- In one embodiment, a computer program product includes a computer readable storage medium having program instructions embodied therewith. The embodied program instructions are executable by a processor to cause the processor to receive, by the processor, a first job request, and analyze, by the processor, the first job request to determine: an estimated complexity of the first job request based on one or more attributes of the first job request and a user skill level of a user that submitted the first job request. Moreover, the embodied program instructions are executable by the processor to admit, by the processor, the first job request to a data analytics system and/or a data storage system in a specified order with respect to other received job requests based on at least: the estimated complexity of the first job request, and the user skill level of the user that submitted the first job request.......................Please see Fig. 1. Abstract.
MASCHMEYER et al. (US 20240311192 A1)- Methods and systems for indicating a resource usage parameter for prompting a large language model (LLM) are described. A user input is received, from an electronic device, for generating a prompt to a LLM. A prompt resource usage parameter is computed based on the user input. A trained resource prediction model is used to generate a predicted response resource usage parameter for a response from the LLM, based on the user input. A total resource usage parameter is computed, based on the prompt resource usage parameter and the predicted response resource usage parameter. A representation of the total resource usage parameter is communicated to the electronic device, to cause the electronic device to provide an output of the representation of the total resource usage parameter.......................Please see Fig. 1. Abstract.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LUCIUS C.G. ALLEN whose telephone number is (703)756-5987. The examiner can normally be reached Mon - Fri 8-5pm (EST).
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, Chineyere Wills-Burns can be reached at (571)272-9752. 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.
/LUCIUS CAMERON GREEN ALLEN/Examiner, Art Unit 2673
/CHINEYERE WILLS-BURNS/Supervisory Patent Examiner, Art Unit 2673
Prosecution Insights