Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
2. This action is in response to the original filing on 04/24/2024. Claims 1-20 are pending and have been considered below.
Information Disclosure Statement
3. The information disclosure statement (IDS(s)) submitted on 05/10/2024 is/are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
4. 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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to the abstract idea without significantly more.
Step 1, the claims are directed to a process, machine, and manufacture.
Step 2A Prong 1, Claims 1, 11, and 20 recite, in part
identifying, a plurality of cached query-response pairs whose queries are similar to that of the particular query (Mental process, human evaluation/judgement concerning similarity).
assign joint probabilities between the particular query and responses from the plurality of cached query-response pairs (Mathematical concepts, calculating and assigning probability relationships).
Step 2A Prong 2, this judicial exception is not integrated into a practical application.
The additional elements:
a device; one or more network interfaces; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process when executed configured to (mere instructions to apply the exception using a generic computer component).
using, by the device, a verification model (mere instructions to apply the exception using a generic computer component).
obtaining, by a device, a particular query for input to a language model; (mere data gathering and recited at a high level of generality, and thus are insignificant extra-solution activity).
providing, by the device, a particular response from the plurality of cached query-response pairs, in lieu of using the particular query as input to the language model to generate a new response (mere data gathering and output recited at a high level of generality, and thus are insignificant extra-solution activity).
Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception, either alone or in combination.
The additional elements:
a device; one or more network interfaces; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process when executed configured to (mere instructions to apply the exception using a generic computer component).
using, by the device, a verification model (mere instructions to apply the exception using a generic computer component).
obtaining, by a device, a particular query for input to a language model; (mere data gathering and recited at a high level of generality, and thus are insignificant extra-solution activity).
providing, by the device, a particular response from the plurality of cached query-response pairs, in lieu of using the particular query as input to the language model to generate a new response (mere data gathering and output recited at a high level of generality, and thus are insignificant extra-solution activity).
Claims 2-10 and 12-19 provide further limitations to the abstract idea (Mathematical concepts and/or Mental processes) as rejected in claims 1, 11, 20, however, they do not disclose any additional elements that would amount to a practical application or significantly more than an abstract idea (data gathering/insignificant extra-solution activity and/or generic computer component).
Claim Rejections – 35 USC § 103
5. 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.
6. Claims 1-5, 8, 9, 11-15, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Muschett et al. (U.S. Patent Application Pub. No. US 20250258847 A1) in view of Wang et al. (U.S. Patent Application Pub. No. US 20210232773 A1).
Claim 1: Muschett teaches a method (i.e. A system and method for improving computer functionality by retrieving answers/responses to questions/input such as those used with chatbots and generative AI systems; para. [0006]) comprising:
obtaining, by a device, a particular query for input to a language model (i.e. FIG. 2 illustrates a high-level example 200 of a client console 102 sending a query 104 such as “How Tall is the Eiffel Tower?” to a generative AI system 108 with a LLM 106 and a LLM cache 220, as shown; para. [0094, 0139]);
identifying, by the device, a plurality of cached query-response pairs whose queries are similar to that of the particular query (i.e. LLM caching of previous questions and responses results in sub-second times. LLM caching of similar questions improves the hit rate over traditional exact matches … A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0097, 0121, 0141, 0148]);
using, by the device, a verification model to, joint probabilities between the particular query and responses from the plurality of cached query-response pairs (i.e. Confidence Score: A numerical value indicating the system's level of confidence in the correctness or relevance of the provided answer. This score helps users and developers gauge the reliability of the response … Ranking or Relevance Score: In cases where multiple potential answers are generated, a ranking or relevance score helps prioritize and present the most suitable answers first. This score is often based on factors like accuracy, recency, and overall relevance; para. [0044, 0047, 0124-0128]); and
providing, by the device and based on the joint probabilities, a particular response from the plurality of cached query-response pairs (i.e. In step 434, in response to the CRS 619 being above a settable value 624, the identified answer from the answer portion of the cache is returned as the answer to QT. This represents a “cache hit”, and the answer is returned or “get” to the question 526; para. [0141, 0146, 0147]), in lieu of using the particular query as input to the language model to generate a new response (i.e. A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0121, 0141, 0147]).
Muschett does not explicitly teach a model to assign joint probabilities.
However, Wang teaches using, by the device, a verification model (i.e. a unified visual-dialogue transformer-based approach or model that leverages a Bidirectional Encoder Representations from Transformers (BERT) pre-trained language models for visual dialogue tasks; para. [0017, 0018]) to assign joint probabilities (i.e. the unified transformer encoder 450 aims to predict whether the appended answer candidate Ât is correct or not based on the joint understanding of vision and previous dialogue history:
PNG
media_image1.png
38
25
media_image1.png
Greyscale
NSP =−E (l,w)˜D log P(y|S(I,w)) Eq (6), where y∈{0,1} indicates whether Ât is correct, and S(⋅) is a binary classifier to predict the probability based on the [CLS] representation T[CLS] at the final layer … the answer candidates are ranked based on their log-likelihood scores; para. [0051-0056]) between the particular query and responses from the plurality of query-response pairs (i.e. The structure 400 employs a unified transformer encoder 450 to encode both the image input 410 and the multi-turn dialogue history input 420. The structure 400 employs the input sequence module 440 to first encode the image input 410 (including caption) and the multi-turn dialogue history input 420 concurrently with the unified transformer encoder 450. In some embodiments, the structure 400 can append an answer option to the input (e.g., appended to the user question 430) and model their interaction in an early fusion manner; para. [0041, 0041, 0047]); and providing, by the device and based on the joint probabilities, a particular response (i.e. The ranking module 334 may rank multiple answer candidates produced by the encoder 332 and provide the ranked answer candidates to the prioritized answer candidate list module 338. The prioritized answer candidate list module 338 may include dense annotations for images that accompany the answer candidates for fine-tuning the ranked answer candidates. After fine tuning, the answer candidate can be selected and provided as a selected answer 356 via the response module 339; para. [0035, 0037, 0077]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 2: Muschett and Wang teach the method as in claim 1. Muschett further teaches comprising: a joint probability between the particular query and a response from each of the plurality of cached query-response pairs based on a likelihood of the response (i.e. Confidence Score: A numerical value indicating the system's level of confidence in the correctness or relevance of the provided answer. This score helps users and developers gauge the reliability of the response … Ranking or Relevance Score: In cases where multiple potential answers are generated, a ranking or relevance score helps prioritize and present the most suitable answers first. This score is often based on factors like accuracy, recency, and overall relevance; para. [0044, 0047, 0124-0128]) being a likely response to the particular query (i.e. LLM caching of previous questions and responses results in sub-second times. LLM caching of similar questions improves the hit rate over traditional exact matches … A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0097, 0121, 0141, 0148]).
Muschett does not explicitly teach assigning a joint probability.
However, Wang further teaches assigning a joint probability between the particular query and a response from each of the plurality of query-response pairs based on a likelihood of the response being a likely response to the particular query (i.e. the unified transformer encoder 450 aims to predict whether the appended answer candidate Ât is correct or not based on the joint understanding of vision and previous dialogue history:
PNG
media_image1.png
38
25
media_image1.png
Greyscale
NSP =−E (l,w)˜D log P(y|S(I,w)) Eq (6), where y∈{0,1} indicates whether Ât is correct, and S(⋅) is a binary classifier to predict the probability based on the [CLS] representation T[CLS] at the final layer … the answer candidates are ranked based on their log-likelihood scores; para. [0051-0056]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 3: Muschett and Wang teach the method as in claim 1. Muschett further teaches comprising: from the plurality of cached query-response pairs as a match or a mismatch (i.e. LLM caching of previous questions and responses results in sub-second times. LLM caching of similar questions improves the hit rate over traditional exact matches … A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0097, 0121, 0141, 0148]).
Muschett does not explicitly teach classifying each response.
However, Wang further teaches classifying each response from the plurality of query-response pairs as a match or a mismatch (i.e. The subject technology can transform the task of selecting an answer from a candidate pool into a pointwise binary classification operation. Concretely, the subject technology can sample an answer Ât from the candidate pool and append it to the input sequence, and cause the NSP head to utilize the final hidden vector T[CLS] of the “[CLS]” as the NSP 460 head for binary classification to distinguish whether the appended answer is correct or not; para. [0051-0056]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 4: Muschett and Wang teach the method as in claim 1. Muschett further teaches from the plurality of cached query-response pairs as a match or a mismatch (i.e. LLM caching of previous questions and responses results in sub-second times. LLM caching of similar questions improves the hit rate over traditional exact matches … A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0097, 0121, 0141, 0148]).
Muschett does not explicitly teach training another language model to classifying each response.
However, Wang further teaches training another language model (i.e. a unified visual-dialogue transformer-based approach or model that leverages a Bidirectional Encoder Representations from Transformers (BERT) pre-trained language models for visual dialogue tasks; para. [0017, 0018]) to classifying each response from the plurality of query-response pairs as a match or a mismatch (i.e. The subject technology can transform the task of selecting an answer from a candidate pool into a pointwise binary classification operation. Concretely, the subject technology can sample an answer Ât from the candidate pool and append it to the input sequence, and cause the NSP head to utilize the final hidden vector T[CLS] of the “[CLS]” as the NSP 460 head for binary classification to distinguish whether the appended answer is correct or not; para. [0051-0056]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 5: Muschett and Wang teach the method as in claim 1. Muschett further teaches providing the particular response to a user (i.e. LLM caching of previous questions and responses results in sub-second times. LLM caching of similar questions improves the hit rate over traditional exact matches … A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0097, 0121, 0141, 0148]); receiving a feedback from the user (i.e. Feedback History: Information about user feedback on the accuracy or helpfulness of the provided answer. This history can be used to continuously improve the system through user feedback loops; para. [0048]), the feedback comprising whether the particular response is a match or a mismatch to the particular query (i.e. Feedback History: Information about user feedback on the accuracy or helpfulness of the provided answer. This history can be used to continuously improve the system through user feedback loops; para. [0048]); and training, based on the feedback, language model to each response from the plurality of cached query-response pairs as a match or a mismatch (i.e. Learning and Adaptation: Incorporates mechanisms for continuous learning and adaptation. This involves updating the generative models with new data to stay current with language trends, user preferences, and evolving conversational patterns; para. [0111]).
Muschett does not explicitly teach another language model to classify each response.
However, Wang further teaches training, based on the feedback, another language model (i.e. The subject technology can initialize the unified transformer encoder with BERT for increased leveraging of the pre-trained language representation. To deeply fuse features from the two modalities, the subject technology make use of two visually-grounded pretraining objectives, such as Masked Language Modeling (MLM) and Next Sentence Prediction (NSP), to train the model on the visual dialogue data; para. [0017, 0018, 0056]) to classify each response from the plurality of query-response pairs as a match or a mismatch (i.e. The subject technology can transform the task of selecting an answer from a candidate pool into a pointwise binary classification operation. Concretely, the subject technology can sample an answer Ât from the candidate pool and append it to the input sequence, and cause the NSP head to utilize the final hidden vector T[CLS] of the “[CLS]” as the NSP 460 head for binary classification to distinguish whether the appended answer is correct or not; para. [0051-0056]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 8: Muschett and Wang teach the method as in claim 1. Muschett further teaches wherein the joint probabilities between the particular query and the responses from the plurality of cached query-response pairs are based the particular query and the responses (i.e. Confidence Score: A numerical value indicating the system's level of confidence in the correctness or relevance of the provided answer. This score helps users and developers gauge the reliability of the response … Ranking or Relevance Score: In cases where multiple potential answers are generated, a ranking or relevance score helps prioritize and present the most suitable answers first. This score is often based on factors like accuracy, recency, and overall relevance; para. [0044, 0047, 0124-0128]).
Muschett does not explicitly teach wherein the joint probabilities are assigned based on comparing different continuations of the particular query and the responses.
However, Wang further teaches wherein the joint probabilities between the particular query and the responses from the plurality of cached query-response pairs are assigned based on comparing different continuations of the particular query and the responses (i.e. To preserve the autoregressive property of the answer generation, the unified transformer encoder 450 can employ the sequence-to-sequence (seq2seq) self-attention mask for a generative setting 482. Namely, the unified transformer encoder 450 may prevent attending future tokens when predicting each token of the answer sequence. Specifically, the structure 400 can divide the input sequence for each transformer block into two subsequences: a previous context sequence (including image input 410 and caption (depicted as I), dialogue history 420 (depicted as He), and question 430 (depicted as Qt)) and the answer sequence: x
PNG
media_image2.png
33
25
media_image2.png
Greyscale
(I,w)=(I,H t ,Q t , t) Eq. (7) … the answer candidates are ranked based on their log-likelihood scores (e.g., 490); para. [0052-0054]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 9: Muschett and Wang teach the method as in claim 1. Muschett further teaches wherein the particular response is associated with a highest joint (i.e. In step 432, using the SLSs 518 (in order of SRV from highest to lowest from layer 1 410 and LLS 630 (in order of LRV from highest to lowest from layer 2 420 and applying a reciprocal rank fusion algorithm to compute a combined ranking set (CRS) for SAT and LAT to provide an identified answer. The process continues to step 434; para. [0010, 0011, 0146]).
However, Wang further teaches wherein the particular response is associated with a highest joint probability (i.e. the answer candidates are ranked based on their log-likelihood scores (e.g., 490) … To fine-tune on these dense annotations, the subject technology combine the NSP scores for all answer candidates together and normalize them into a probability distribution [p1, . . . , p100]. In other aspects, a cross entropy (CE) loss L can be employed to optimize the model; para. [0052-0056]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
Claim 11 is similar in scope to Claim 1 and is rejected under a similar rationale.
Muschett teaches an apparatus, comprising: one or more network interfaces; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process when executed configured to (i.e. one or more processors 1402 or processing units, a system memory 1410, and a bus 1430 that couples various system components, including the system memory to the processor … the information processing system can communicate with one or more networks such as a local area network (LAN), a general wide area network (WAN), and/or a public network (e.g., the Internet) 1462 via network adapter 1460 to LLM 1464. As depicted, the network adapter communicates with the other components of information processing system via the bus; para. [0166, 0170]).
Claims 12-15, 18, and 19 are similar in scope to Claims 2-5, 8, 9 and are rejected under a similar rationale.
Claim 20 is similar in scope to Claim 1 and is rejected under a similar rationale.
Muschett teaches a tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising (i.e. a storage system can be provided for reading from and writing to a non-removable or removable, non-volatile media such as one or more solid state disks and/or magnetic media (typically called a “hard drive”) 1416. A magnetic disk drive for reading from and writing to a removable, non-volatile magnetic disk (e.g., a “floppy disk”), and an optical disk drive for reading from or writing to a removable, non-volatile optical disk such as a CD-ROM, DVD-ROM or other optical media can be provided. In such instances, each can be connected to the bus 1430 by one or more data media interfaces 1460. The memory 1410 can include at least one program product having a set of program modules that are configured to carry out the functions of an embodiment of the present disclosure; para. [0168]).
7. Claims 6 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Muschett in view of Wang, and further in view of Dang et al. (U.S. Patent Application Pub. No. US 20240070489 A1).
Claim 6: Muschett and Wang teach the method as in claim 1. Muschett further teaches wherein the plurality of cached query-response pairs (i.e. A natural language query (the user request) and its corresponding LLM answer or response will be cached. When a subsequent user request is received, the cache is first checked to see if a similar request has already been made and, if so will respond with the corresponding LLM answer from the cache; eliminating the downstream request to engineer a prompt, generate a context window and subsequently query the LLM; para. [0121]) comprises a predetermined number of top match to the particular query (i.e. In all layers, it is important to understand the confidence score or relevance value of the top result (k=1) from the retriever. A confidence score (CS) at each layer will be established on a use-case-by-use-case basis. The confidence score for. Layer-1 is referred to as semantic relevance values (SRV) in a semantic layer set (SLS) of semantic answers (SAT) with associated answer-metadata. Layer 2 is referred to as lexical relevance values (LRV) in a lexical layer set (LLS) of lexical answers (LAT). Layer 3 is referred to as a combined set (CRS) for SAT and LAT … In step 432, using the SLSs 518 (in order of SRV from highest to lowest from layer 1 410 and LLS 630 (in order of LRV from highest to lowest from layer 2 420 and applying a reciprocal rank fusion algorithm to compute a combined ranking set (CRS) for SAT and LAT to provide an identified answer; para. [0124-0128, 0146]).
Muschett does not explicitly teach a predetermined number of top matches.
However, Dang teaches wherein the plurality of cached query-response pairs comprises a predetermined number of top matches to the particular query (i.e. method 200 can compute similarities between a query embedding and the question embeddings and select the top N question embeddings based on the similarity scores. Method 200 can then select the top M answers for each of the N questions; para. [0058]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Muschett and Wang to include the feature of Dang. One would have been motivated to make this modification because it provides a fixed-size set of the most relevant query response candidates for further verification and improving retrieval consistency and response relevance.
Claim 16 is similar in scope to Claim 6 and is rejected under a similar rationale.
8. Claims 7 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Muschett in view of Wang, and further in view of Kim et al. (U.S. Patent Application Pub. No. US 20240311405 A1).
Claim 7: Muschett and Wang teach the method as in claim 1. Muschett further teaches wherein the verification model comprises the language model (i.e. the proposed solution provides a solution to mitigate GenAI processing delays and costs by introducing an LLM cache as shown in FIG. 2; para. [0096]).
Muschett does not explicitly teach wherein the model comprises a second language model, wherein the second language model is smaller than the language model
However, Wang further teaches wherein the verification model comprises a second language model (i.e. a unified visual-dialogue transformer-based approach or model that leverages a Bidirectional Encoder Representations from Transformers (BERT) pre-trained language models for visual dialogue tasks; para. [0017, 0018, 0051, 0055]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the invention of Muschett to include the feature of Wang. One would have been motivated to make this modification because it improves the accuracy and reliability of the cached response selection.
However, Kim teaches wherein the verification model comprises a second language model, wherein the second language model is smaller than the language model (i.e. the candidate generative models include a larger LLM that includes over 200 billion parameters and a smaller LLM that includes twenty, thirty, forty, fifty, or other percent less parameters than the larger LLM. For instance, the smaller LLM can include less than 100 billion parameters. In some implementations, the smaller LLM can be a quantized and/or pruned version of the larger LLM; para. [0005, 0006]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Muschett and Wang to include the feature of Kim. One would have been motivated to make this modification because the smaller LLM can be more computationally efficient than the larger LLM.
Claim 17 is similar in scope to Claim 7 and is rejected under a similar rationale.
9. Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Muschett in view of Wang, and further in view of Arora et al. (U.S. Patent Application Pub. No. US 20200175971 A1).
Claim 10: Muschett and Wang teach the method as in claim 1. Muschett does not explicitly teach wherein the particular query comprises a status of a device in a computer network.
However, Arora teaches wherein the particular query comprises a status of a device in a computer network (i.e. Examples of the voice query 103 can include questions relating to the satellite communication system. For instance, the voice query 103 can include a question about a present condition of the system, a performance of a particular subsystem, a status of a particular component, or another statistic or property of the system. The query 103 can relate to a previous condition, status, performance metric, statistic, or property of the system (e.g., requesting the average hourly data traffic during the previous 24 hours); para. [0032, 0034, 0037]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to modify the combination of Muschett and Wang to include the feature of Arora. One would have been motivated to make this modification because it provides a known network management query environment, where users submit natural language queries regarding the status, condition, or performance of network components.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
Lovric (Pub. No. US 20250165752 A1), the techniques described herein relate to a method including: receiving a query; determining that the query matches a cached query; retrieving a cached response from a large language model provider for the cached query; and providing the cached response.
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TAN TRAN whose telephone number is (303)297-4266. The examiner can normally be reached on Monday - Thursday - 8:00 am - 5:00 pm MT.
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, Matt Ell can be reached on 571-270-3264. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/TAN H TRAN/Primary Examiner, Art Unit 2141
Prosecution Insights