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 .
Drawings
The drawings are objected to as failing to comply with 37 CFR 1.84(p)(5) because they do not include the following reference sign(s) mentioned in the description:
“method 400” (see e.g. par. [0043]);
“method 600” (see e.g. par. [0057]);
“four ICs 700[B]-D” (see e.g. par. [0078]).
Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Objections
Claims 6 and 19 are objected to because of the following informalities:
Claim 6 recites “the IRs each comprises values”. It is believed this should read “the IRs each comprise values”.
Claim 19 recites “the models each comprises values”. It is believed this should read “the models each comprise values”.
Appropriate correction is required.
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 (i.e., changing from AIA to pre-AIA ) 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, 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.
Claim(s) 1-3, 7-9, 11-13 and 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2023/0305818 to Lichtenau et al. (Lichtenau) in view of US 2022/0366267 to Karmakar et al. (Karmakar).
Claims 1, 11 and 17: Lichtenau discloses a method, comprising:
receiving a template identifying a base artificial intelligence (AI) model (par. [0025] “AI accelerator code … that includes variables”) and structural parameters corresponding to a second AI model (par. [0025] “User inputs (e.g., tensor, tensor size, AI operation, etc.)”), wherein the base AI model has been previously compiled to execute on a model-specific chipset (par. [0029] “machine code”);
modifying compilation data corresponding to the base AI model using the structural parameters (par. [0025] “variables are then looked up … and are inserted in to the final accelerator code”);
creating, at a compiler, executable code for the second AI model using the modified compilation data (par. [0022] “compile special code for an AI accelerator”); and
executing the second AI model on the model-specific chipset using the executable code (par. [0039] “execute the variable replaced AI accelerator code”).
Lichtenau does not explicitly disclose receiving a template file.
Karmakar teaches receiving a template file (par. [0037] “example input formats include … MxNet (.json)”).
It would have been obvious before the effective filing date of the claimed invention to provide the template as a file. Those of ordinary skill in the art would have been motivated to do so as a known means of representing AI models to AI accelerators which would have produced only the expected result.
Claims 2 and 12: Lichtenau and Karmakar teach The method of claim 1, 11, further comprising, after receiving the template file but before modifying the compilation data:
identifying code corresponding to the base AI model stored in a library (par. [0025] “providing a number of high-level AI primitives to a user”, par. [0034] “accelerator code … can be fetched by DMA interface 11”), wherein modifying the compilation data comprises:
replacing values in the code for the base AI model with values in the structural parameters (par. [0025] “variables are then looked up … and are inserted in to the final accelerator code”).
Claims 3, 13 and 18: Lichtenau and Karmakar teach claims 2, 12 and 17, wherein the library stores code for a plurality of base AI models that are supported by the compiler (par. [0025] “providing a number of high-level AI primitives to a user”).
Lichtenau does not explicitly disclose wherein the plurality of base AI models have each been previously executed on the model-specific chipset.
It would have been obvious before the effective filing date of the claimed invention to reuse the base AI models. Those of ordinary skill in the art would have been motivated to do so to avoid having to re-develop them for each use.
Claim 7: Lichtenau and Karmakar teach the method of claim 1, wherein the template file does not include software code for a programming language (par. [0025] “User inputs (e.g., tensor, tensor size, AI operation, etc.)”, Karmakar par. [0037] “MxNet (.json)”).
Claim 8: Lichtenau and Karmakar teach claim 7, wherein data in the template file uses a JavaScript Object Notation (JSON) format (Karmakar par. [0037] “MxNet (.json)”).
Claim 9: Lichtenau and Karmakar teach claim 1, wherein structural parameters change an architecture or structure of the base AI model to convert the base AI model into the second AI model (Lichtenau par. [0025] “tensor, tensor size, AI operation, etc.”).
Claim(s) 4-6, 14-15 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2023/0305818 to Lichtenau et al. (Lichtenau) in view of US 2022/0366267 to Karmakar et al. (Karmakar) in view of US 12,046,028 to Wu et al. (Wu).
Claims 4 and 14: Lichtenau and Karmakar teach claims 1 and 11, further comprising, after receiving the template file but before modifying the compilation data:
identifying the base AI model that is stored in a library (par. [0025] “AI primitives”), wherein modifying the compilation data comprises:
replacing values in the base AI model with values in the structural parameters to generate a modified IR (par. [0025] “variables are then … inserted in to the final accelerator code”).
Lichtenau does not explicitly disclose:
an intermediate representation corresponding to the base AI model.
Wu teaches:
an intermediate representation corresponding to an AI model (col. 6, lines 38-41 “the IR Graph corresponding to the CNN model”).
It would have been obvious before the effective filing date of the claimed invention to identify an IR corresponding to the base AI model. Those of ordinary skill in the art would have been motivated to do so as an alternate method of representing accelerator code which would have produced only the expected results.
Claims 5 and 15: Lichtenau and Wu teach claims 4 and 15, wherein the modified IR has values for various arguments to perform operations that are part of the second AI model (Lichtenau par. [0025] “providing a number of high-level AI primitives to a user”).
Claim 6: Lichtenau and Wu teach the method of claim 4, wherein the library stores IRs for a plurality of base AI models that are supported by the compiler (Lichtenau par. [0025] “providing a number of high-level AI primitives to a user”, Wu col. 6, lines 38-41 “the IR Graph corresponding to the CNN model”), wherein the IRs each comprises values of arguments that configure the model-specific chipset to perform operations for a respective base AI model (Lichtenau par. [0025] “template AI accelerator code … that includes variables”).
Lichtenau does not explicitly disclose wherein the plurality of base AI models have each been previously executed on the model-specific chipset.
It would have been obvious before the effective filing date of the claimed invention to reuse the base AI models. Those of ordinary skill in the art would have been motivated to do so to avoid having to re-develop them for each use.
Claim 19: Lichtenau and Karmakar teach the system of claim 17, wherein the operation further comprises, after receiving the template file but before modifying the compilation data:
identifying the base AI model that is stored in a library,
wherein modifying the compilation data comprises:
replacing values in the base AI model with values in the structural parameters to generate a modified IR (par. [0025] “variables are then … inserted in to the final accelerator code”),
wherein the modified base AI model has values for various arguments to perform operations that are part of the second AI model (par. [0039] “the variable replaced AI accelerator code”), and
wherein the library stores models for a plurality of base AI models that are supported by the compiler, wherein the models each comprises values of arguments that configure the model-specific chipset to perform operations for a respective base AI model (par. [0039] “execute the variable replaced AI accelerator code”).
Lichtenau does not explicitly disclose wherein the plurality of base AI models have each been previously executed on the model-specific chipset.
It would have been obvious before the effective filing date of the claimed invention to reuse the base AI models. Those of ordinary skill in the art would have been motivated to do so to avoid having to re-develop them for each use.
Lichtenau does not explicitly disclose:
an intermediate representation corresponding to the base AI model.
Wu teaches:
an intermediate representation corresponding to an AI model (col. 6, lines 38-41 “the IR Graph corresponding to the CNN model”).
It would have been obvious before the effective filing date of the claimed invention to identify an IR corresponding to the base AI model. Those of ordinary skill in the art would have been motivated to do so as an alternate method of representing accelerator code which would have produced only the expected results.
Claim(s) 10, 16 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2023/0305818 to Lichtenau et al. (Lichtenau) in view of US 2022/0366267 to Karmakar et al. (Karmakar) in view of US 12,524,266 to Yu et al. (Yu).
Claims 10, 16 and 20: Lichtenau and Karmakar teach claims 1 and 11, but do not explicitly teach wherein the base AI model and the second AI model are different types of transformer models, wherein the model-specific chipset can is optimized to execute only transformer models.
Yu teaches:
transformer models and a model-specific chipset optimized for transformer models (col. 5, lines 3-16 “tensor processing units (TPUs) … optimized to perform those operations”).
It would have been obvious before the effective filing date of the claimed invention to use transformer models and a transformer optimized chipset. Those of ordinary skill in the art would have been motivated to do so as a known type of AI model/chipset which would have produced only the expected results (e.g. optimization for a specific purpose of the model).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JASON D MITCHELL whose telephone number is (571)272-3728. The examiner can normally be reached Monday through Thursday 7:00am - 4:30pm and alternate Fridays 7:00am 3:30pm.
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, Lewis Bullock can be reached at (571)272-3759. 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.
/JASON D MITCHELL/Primary Examiner, Art Unit 2199