DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1, 3, 4, 10, 12, 14, and 20 have been amended.
Claims 2, 8, 9, 13, and 19 have been cancelled.
Claims 1, 3-7, 10-12, 14-18, and 20 have been examined.
The specification and drawing objections in the previous Office Action have been addressed and are withdrawn, except as otherwise indicated below.
The § 112 rejections in the previous Office Action have been addressed and are withdrawn.
Specification
The disclosure is objected to because of the following informalities.
The title of the invention is not descriptive. A new title is required that is clearly indicative of the invention to which the claims are directed.
Appropriate correction is required. The lengthy specification has not been checked to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Examiner’s Note
Claims 10 and 11 are improperly ordered. The claims fail to comply with MPEP § 608.01(n)(IV), which states, “A claim which depends from a dependent claim should not be separated therefrom by any claim which does not also depend from said "dependent claim."” Claim 10 depends from dependent claim 3, and is separated therefrom by claims 4-7, which do not depend from claim 3. The Examiner can correct the claim ordering if/when the application is in condition for allowance. In the future, please follow the guidelines above to ensure correct claim ordering. Doing so helps to promote compact prosecution.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1, 3-7, 10-12, 14-18, and 20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor, or for pre-AIA the applicant regards as the invention.
Claim 1 recites, “for each processor.” There is insufficient antecedent basis for this limitation in the claim. It is unclear whether this refers to each processor of the previously introduced “at least one processor,” or “plurality of processors.” This ambiguity renders the scope of the claims indefinite. For purposes of examination, the latter interpretation is taken. Claims 12 and 20 include similar limitations and are similarly rejected.
Claim 1 recites, “the processor.” There is insufficient antecedent basis for this limitation in the claim. For purposes of examination, this limitation is interpreted as, “the each processor.” Claims 12 and 20 include similar limitations and are similarly rejected.
Claims 3-7, 10, 11, and 14-18 are rejected as depending from rejected base claims and failing to cure the indefiniteness of those base claims.
Claim Rejections - 35 USC § 103
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, 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 1, 3-7, 10-12, 14-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Non-Patent Literature “LoRA: Low-Rank Adaptation of Large Language Models” by Hu et al. (hereinafter referred to as Hu) in view of US Publication No. 2024/0111528 by Tan et al. (previously cited and hereinafter referred to as “Tan”).
Regarding claims 1, 12, and 20, Hu discloses:
a task execution method for a large model, comprising: executing, using a target feature to be processed as an input, a collaborative computing task using at least one processor to obtain a target collaborative feature, wherein the collaborative computing task comprises a first collaborative task and a second collaborative task, the first collaborative task is configured to process the target feature to be processed and a first collaborative sub-weight to obtain an intermediate collaborative feature, the second collaborative task is configured to process the intermediate collaborative feature and a second collaborative sub-weight to obtain the target collaborative feature, and the first collaborative sub-weight and the second collaborative sub-weight are determined by decomposing a collaborative weight into a product of two matrices according to a general matrix multiplication mechanism, wherein a dimension of the first collaborative sub- weight and a dimension of the second collaborative sub-weight are both smaller than a dimension of the collaborative weight (Hu discloses, at Figure 1 and § 3, multiplying sub-matrix A (first collaborative sub-weight) by x (target feature) to produce an intermediate result and then multiplying the intermediate result with sub-matrix B (second collaborative sub-weight) to obtain a result. The sub-matrices represent a decomposition product (BA) and each have a dimension r that is smaller than the original dimension d. Hu also discloses, at § 3.1, a GPU implementation, which discloses processors, memory, and instructions in media implementations.); and
fusing a target basic feature and the target collaborative feature to obtain a next target feature to be processed, wherein the target basic feature is obtained by executing a basic computing task using the at least one processor, and the basic computing task is configured to process a basic weight and the target feature to be processed (Hu discloses, at Figure 1 and § 3, adding (fusing) the results of the low rank computations (collaborative task) with the pre-trained result (basic computing task with basic weight and target feature) to produce a result which is used in subsequent iterations.),
wherein the basic computing task comprises a computing task executed by a base model including the basic weight, the collaborative computing task comprises a computing task executed by a collaborative model including the collaborative weight, and the collaborative computing task is a low-rank adaptation computing task (Hu discloses, at Figure 1 and § 3, a pre-trained weight matrix W (base model including the basic weight and a low rank decomposition BA (collaborative computing task executed by a collaborative model that is a low-rank adaptation task).),
wherein the first collaborative task comprises a plurality of first collaborative sub- tasks, and the plurality of first collaborative sub-tasks are executed… (Hu discloses, at § 3, performing matrix multiplication on input and weight sub-matrices, which is understood to be an iterative process including multiple sub-tasks.);
wherein executing the collaborative computing task using the at least one processor comprises: for each processor, reading the first collaborative sub-weight and a target sub- feature to be processed corresponding to a first collaborative sub-task assigned to the processor from a memory using the processor, wherein the target sub-feature to be processed is selected from a plurality of target sub-features to be processed that are obtained by partitioning the target feature to be processed according to a predetermined partitioning strategy (Hu discloses, at § 3, performing computations using weight and input values, which discloses reading the values from memory, wherein the sub-matrices are obtained by partitioning the original weight and input matrices, which discloses a predetermined partitioning strategy.); and
executing the first collaborative sub-task using the at least one processor based on the first collaborative sub-weight and the target sub-feature to be processed to obtain a first intermediate collaborative sub-feature, wherein the intermediate collaborative feature is determined according to first intermediate collaborative sub-features corresponding to the plurality of first collaborative sub-tasks respectively (Hu discloses, at § 3, performing matrix multiplication using low rank weight sub-matrix A and corresponding input values, which discloses first collaborative sub-tasks and intermediate collaborative sub-features.).
Hu does not explicitly disclose the aforementioned collaborative sub-tasks are executed in parallel by a plurality of processors.
However, in the same field of endeavor (e.g., machine learning) Tan discloses:
processing on multiple processors in parallel (Tan discloses, at Figure 2 and related description, a plurality of vector compute engines (processors). See, e.g. ¶ [0029]. The vector compute engines operate in parallel on submatrices. See, e.g., ¶ [0034] et seq.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify Hu to include parallel processors, as disclosed by Tan, in order to improve performance by enabling parallel execution.
Regarding claims 3 and 14, taking claim 3 as representative, Hu discloses the elements of claim 1, as discussed above. Hu also discloses:
the plurality of first collaborative sub- tasks are associated with the same first collaborative sub-weight (Hu discloses, at § 3, performing matrix multiplication, which discloses iterative computations for a given matrix.).
Regarding claims 4 and 15, taking claim 4 as representative, Hu discloses the elements of claim 1, as discussed above. Hu also discloses:
wherein the second collaborative task comprises a plurality of second collaborative sub-tasks (Hu discloses, at § 3, performing matrix multiplication on intermediate result and weight sub-matrices, which is understood to be an iterative process including multiple sub-tasks.);
wherein executing the collaborative computing task using the at least one processor further comprises: reading a second intermediate collaborative sub-feature corresponding to a second collaborative sub-task from a memory using the at least one processor, wherein the second intermediate collaborative sub-feature is determined according to the intermediate collaborative feature (Hu discloses, at § 3, performing computations using weight and intermediate values, which discloses reading the values from memory, wherein the sub-matrices are obtained by partitioning (dividing) the original weight and input matrices.); and
executing the second collaborative sub-task using the at least one processor based on the second intermediate collaborative sub-feature and the second collaborative sub-weight to obtain a target collaborative sub-feature, wherein the target collaborative feature is determined according to target collaborative sub-features corresponding to the plurality of second collaborative sub-tasks respectively (Hu discloses, at § 3, performing matrix multiplication using low rank weight sub-matrix B and corresponding intermediate values, which discloses second collaborative sub-tasks.).
Regarding claims 5 and 16, taking claim 5 as representative, Hu discloses the elements of claim 4, as discussed above. Hu also discloses:
the plurality of second collaborative sub-tasks are associated with the same second collaborative sub-weight (Hu discloses, at § 3, performing matrix multiplication, which discloses iterative computations for a given matrix.).
Regarding claims 6 and 17, taking claim 6 as representative, Hu discloses the elements of claim 1, as discussed above. Hu also discloses:
the target feature to be processed is determined according to an initial feature; or the target feature to be processed is obtained by the at least one processor executing a previous collaborative computing task and a previous basic computing task (Hu discloses, at § 3, using an input, which discloses an initial feature or, in the case of successive training iterations, a resulting set of inputs.).
Regarding claims 7 and 18, taking claim 7 as representative, Hu discloses the elements of claim 6, as discussed above. Hu also discloses:
the target feature to be processed comprises a target text feature to be processed, the initial feature is determined according to an initial text, and an execution result of the at least one processor executing the basic computing task and the collaborative computing task is an output text corresponding to the initial text (Hu discloses, at Title, large language models, which discloses text as an input and text as an output.).
Regarding claim 10, Hu discloses the elements of claim 3, as discussed above. Hu also discloses:
the second collaborative task comprises a plurality of second collaborative sub-tasks (Hu discloses, at § 3, performing matrix multiplication on intermediate result and weight sub-matrices, which is understood to be an iterative process including multiple sub-tasks.);
wherein executing the collaborative computing task using the at least one processor further comprises: reading a second intermediate collaborative sub-feature corresponding to a second collaborative sub-task from a memory using the at least one processor, wherein the second intermediate collaborative sub-feature is determined according to the intermediate collaborative feature (Hu discloses, at § 3, performing computations using weight and intermediate values, which discloses reading the values from memory, wherein the sub-matrices are obtained by partitioning (dividing) the original weight and input matrices.); and
executing the second collaborative sub-task using the at least one processor based on the second intermediate collaborative sub-feature and the second collaborative sub-weight to obtain a target collaborative sub-feature, wherein the target collaborative feature is determined according to target collaborative sub-features corresponding to the plurality of second collaborative sub-tasks respectively (Hu discloses, at § 3, performing matrix multiplication using low rank weight sub-matrix B and corresponding intermediate values, which discloses second collaborative sub-tasks.).
Regarding claim 11, Hu discloses the elements of claim 10, as discussed above. Hu also discloses:
the plurality of second collaborative sub-tasks are associated with the same second collaborative sub-weight (Hu discloses, at § 3, performing matrix multiplication, which discloses iterative computations for a given matrix.).
Response to Arguments
On page 11 of the response filed August 12, 2026 (“response”), the Applicant argues, “A new title is submitted herewith that is clearly indicative of the invention to which the claims are directed. It is respectfully requested that the objection be withdrawn.”
Though fully considered, the Examiner respectfully disagrees. The new title is “COLLABORATIVE COMPUTING TASK EXECUTION METHOD FOR LARGE MODEL, ELECTRONIC DEVICE, AND STORAGE MEDIUM.” The title indicates an entire field of invention rather than any particular invention. The Applicant is requested to consider what the inventive aspect captured in the claims is and submit a title that is briefly descriptive of the inventive aspect. Accordingly, the Applicant’s arguments are deemed unpersuasive.
On pages 13-15 of the response the Applicant argues that Hu does not anticipate the amended claims.
These remarks have been fully considered and, in light of the claim amendments presented in the response, are deemed persuasive. Please see above for new grounds of rejection of the amended claims. Specifically, the Applicant amended the claims to add parallel processing of collaborative sub-tasks by a plurality of processors. Parallel processing using a plurality of processors is notoriously well-known and is implicit in Hu’s disclosure of multiple GPUs. See § 3. However, rather than rely on implicit disclosure, the Examiner cites to Tan for explicit disclosure of parallel processing using multiple processors. See, e.g., Figure 2 and related description.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHAWN DOMAN whose telephone number is (571)270-5677. The examiner can normally be reached on Monday through Friday 8:30am-6pm Eastern Time.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jyoti Mehta can be reached on 571-270-3995. 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.
/SHAWN DOMAN/
Primary Examiner, Art Unit 2183
Prosecution Insights