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 .
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-9 and 12-20 are rejected under 35 U.S.C. 103 as being unpatentable over Jsonformer (GitHub repository) “Jsonformer: A Bulletproof Way to Generate Structure JSON from Language Models” by 1rgs (GitHub username), available at: https://github.com/1rgs/jsonformer, Internet Archive Wayback Machine snapshot of May 5, 2023, hereinafter JSONFORMER, in view of Shamir; Gil et al. (US 20250292098 A1) hereinafter SHAMIR
Regarding claim 1, JSONFORMER teaches:
instantiate a trained machine learning model;
As shown in Fig 1. JSONFORMER instantiate a trained machine learning model in the following lines of code:
“from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("databricks/dolly-v2-12b")” (JSONFORMER, Fig. 1).
PNG
media_image1.png
567
625
media_image1.png
Greyscale
Fig. 1 JSONFORMER: README – “Example” section, snapshot of May 5, 2023
instantiate a model plugin that, during inference, is configured to: receive an input sequence of input tokens of a prompt, the prompt including context and a structured output definition;
“prompt = "Generate a person's information based on the following schema:" jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()
print(generated_data)” (JSONFORMER, Fig. 1)
identify one or more deterministic input tokens corresponding to the structured output definition;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
in response to identifying the one or more deterministic input tokens: skip transmission of the one or more deterministic input tokens to the machine learning model;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
and write the one or more deterministic input tokens as one or more deterministic output tokens to an output token sequence;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
“Efficiency: By generating only the content tokens and filling in the fixed tokens, Jsonformer is more efficient than generating a full JSON string and parsing it.” (JSONFORMER, README – “Features” section).
pass a remainder of input tokens in the input sequence other than the one or more deterministic input tokens to the machine learning model;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
and output the output sequence including the one or more deterministic output tokens and the other output tokens generated by the probabilistic token-wise generation.
“prompt = "Generate a person's information based on the following schema:"
jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()
print(generated_data)” (JSONFORMER, Fig. 1).
JSONFORMER does not explicitly disclose, but SHAMIR teaches:
A computing system, comprising: processing circuitry and associated memory storing instructions that when executed cause the processing circuitry to:
“One example aspect of the present disclosure is directed to a computer-implemented method for performing preference optimization, the method comprising: obtaining, by a computing system comprising one or more computing devices, a training tuple comprising a training sequence comprising a sequence of tokens; processing, by the computing system, at least a portion of the sequence of tokens in the training sequence with a sequence processing model to generate base probabilities respectively for one or more candidate tokens included in a token vocabulary; …” (SHAMIR [0007]).
via the machine learning model, perform probabilistic token-wise generation of other output tokens in the output sequence based on the remainder of the input tokens without the one or more deterministic input tokens;
“The sequence processing model 106 and the base prediction layer 116 generate base probabilities 120, which are the probabilities associated with each candidate token from a predetermined token vocabulary succeeding the input portion of the training sequence 102. These base probabilities 120 are a reflection of the model's assessment of the likelihood of each token's occurrence in the natural progression of the training sequence.” (SHAMIR [0091]).
“Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance quality where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution and pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem. ” (SHAMIR [0234]).
“Prediction layer(s) 6 can evaluate associations between portions of input sequence 5 and a particular output element. These associations can inform a prediction of the likelihood that a particular output follows the input context. For example, consider the textual snippet, “The carpenter's toolbox was small and heavy. It was full of ______.” Example prediction layer(s) 6 can identify that “It” refers back to “toolbox” by determining a relationship between the respective embeddings. Example prediction layer(s) 6 can also link “It” to the attributes of the toolbox, such as “small” and “heavy.” Based on these associations, prediction layer(s) 6 can, for instance, assign a higher probability to the word “nails” than to the word “sawdust.” ” (SHAMIR [0203]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified the invention of _ taught by JSONFORMER with the invention of _ taught by SHAMIR to include further details and narrow down the generation of an output by the machine learning model. The motivation to do so is teach by SHAMIR in “[0234] ...In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.” (SHAMIR [0234]).
Regarding claim 2, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 1, wherein the machine learning model is a generative language model;
“model = AutoModelForCausalLM.from_pretrained("databricks/dolly-v2-12b") tokenizer = AutoTokenizer.from_pretrained("databricks/dolly-v2-12b")“ (JSONFORMER, Fig. 1).
Wherein the input argument “databricks/dolly-v2-12b” is a trained Large Language Model.
and the context includes unstructured natural language text.
“prompt = "Generate a person's information based on the following schema:"“ (JSONFORMER, Fig. 1).
PNG
media_image2.png
827
860
media_image2.png
Greyscale
Fig. 2 JSONFORMER: README – “Example” section, snapshot of May 5, 2023
Regarding claim 3, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 1, wherein the structured output definition includes fixed output text interleaved with text generation statements.
See Fig. 2, wherein the structured output is interleaved with text generation statements. Specifically in the instruction shown: ” {"type": "object", "properties": {"car": {"type": "object", "properties": {"make": {"type": "string"}, …”, wherein “car”, “make”, the colons (:) and the brackets ({}) represent the structured part of the text and the “{"type": "string"}” and “{"type": "object"}” represent the text generation statements and the result from that generation is shown in Fig. 2.
PNG
media_image3.png
705
828
media_image3.png
Greyscale
Fig. 3 JSONFORMER: README – “Problem - Solution” section, snapshot of May 5, 2023
Regarding claim 4, the rejection of claim 3 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 3, wherein a preprocessor of the model plugin is configured to sequentially process the interleaved fixed output text and text generation statements, to thereby interleave the deterministic output tokens for the fixed output text and the other output tokens, the other output tokens including probabilistically generated output tokens generated in response to the text generation statements in the output sequence.
JSONFORMER teaches by example, in their sequential code, how the schema of the fixed output tokens is defined with the areas designated for the generation model output specified. Then that parameter is sent to a function with the language model, the tokenizer, and a prompt into a function that would be used to generate the data as shown in the following code:
“json_schema = {
"type": "object",
"properties": {
"name": {"type": "string"},
"age": {"type": "number"},
"is_student": {"type": "boolean"},
"courses": {
"type": "array",
"items": {"type": "string"}
}
}
}
prompt = "Generate a person's information based on the following schema:"
jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()” (JSONFORMER, Fig. 1).
The output format (not the direct output from the previous code) is further disclosed in Fig. 3, wherein the author points out the fix section of the output (in the left side of the image) and then points out the generated content tokens (in the right side of the image).
Furthermore, JSONFORMER disclose: “Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
Regarding claim 5, the rejection of claim 4 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 4, wherein processing the fixed output text is accomplished at least in part by the writing of the one or more deterministic output tokens to the output sequence;
See Fig. 2, wherein the output for the format listed results in the deterministic token, for example, “make” displayed in the output as “make”.
and processing the text generation statements is accomplished by passing the remainder of input tokens to the machine learning model, where the remainder includes input tokens for the context and the text generation statements.
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
Furthermore, JSONFORMER disclose in Fig. 2 how the non-deterministic tokens like “{"type": "string",” near to “make” becomes “audi” after the generative model process the request based on a schema and a prompt.
Regarding claim 6, the rejection of claim 4 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 4, wherein fixed output text and the text generation statements are labeled by respective preprocessor directives that are interpreted by the preprocessor.
See Fig. 2 wherein the output generated has 2 distinctive colors to identify the nature of the origin of the text, for fix/ deterministic text related to the structure the color is white and for the machine generated text the color is blue (in case the image is in black and white, consider the lighter text to be the deterministic structure and the darker text the generated text).
Regarding claim 7, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 1, wherein skipping transmission of the one or more deterministic input tokens is performed by masking or omitting the one or more deterministic input tokens in a modified input sequence that is transmitted to the machine learning model.
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
See Fig. 3, wherein the omitted deterministic text is pointed by JSONFORMER on the left side of the image, including text, braces, semi-colon, and brackets and the right side pointing at the part of the text that was transmitted to the language model to generate content to fill the space.
Regarding claim 8, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 1, wherein the processing circuitry is configured to convert, via a tokenizer, the output sequence into a response including deterministically generated text based on the deterministic output tokens interleaved with probabilistically generated text based on the other output tokens.
See Fig. 1, wherein a tokenizer is defined and used as a parameter for the model that generates the output data, alongside the language model, the deterministic text and the prompt. The tokenizer is described in the following line of code:
“tokenizer = AutoTokenizer.from_pretrained("databricks/dolly-v2-12b")” (JSONFORMER, Fig. 1).
“jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()” (JSONFORMER, Fig. 1).
Regarding claim 9, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
The computing system of claim 1, wherein the structured output definition is defined by a programming language, markup language, domain specific language, context free grammar, regular expression, schema, mathematical notation, or chemical formula.
“Jsonformer works on complex schemas, even with tiny models. Here is an example of a schema with nested objects and arrays, generated by a 3B parameter model.” (JSONFORMER, README – “Example” section).
“Bulletproof JSON generation: Jsonformer ensures that the generated JSON is always syntactically correct and conforms to the specified schema.” (JSONFORMER, README – “Features” section).
Wherein JSONFORMER schema satisfy at least one of the items in the list.
Regarding claim 12, JSONFORMER teaches:
instantiating a trained machine learning model;
As shown in Fig 1. JSONFORMER instantiate a trained machine learning model in the following lines of code:
“from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("databricks/dolly-v2-12b")” (JSONFORMER, Fig. 1)
instantiating a model plugin;
“from jsonformer import Jsonformer” (JSONFORMER, Fig. 1)
“jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)” (JSONFORMER, Fig. 1)
during inference by the trained machine learning model: receiving an input sequence of input tokens of a prompt, the prompt including context and a structured output definition;
“prompt = "Generate a person's information based on the following schema:" jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()
print(generated_data)” (JSONFORMER, Fig. 1)
identifying one or more deterministic input tokens corresponding to the structured output definition;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
in response to identifying the one or more deterministic input tokens: skipping transmission of the one or more deterministic input tokens to the machine learning model;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
and writing the one or more deterministic input tokens as one or more deterministic output tokens to an output token sequence;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
“Efficiency: By generating only the content tokens and filling in the fixed tokens, Jsonformer is more efficient than generating a full JSON string and parsing it.” (JSONFORMER, README – “Features” section).
passing a remainder of input tokens in the input sequence other than the one or more deterministic input tokens to the machine learning model;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
and outputting the output sequence including the one or more deterministic output tokens and the other output tokens generated by the probabilistic token-wise generation.
“prompt = "Generate a person's information based on the following schema:"
jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()
print(generated_data)” (JSONFORMER, Fig. 1).
JSONFORMER does not explicitly disclose, but SHAMIR teaches:
A computerized method, comprising:
“One example aspect of the present disclosure is directed to a computer-implemented method for performing preference optimization, the method comprising: obtaining, by a computing system comprising one or more computing devices, a training tuple comprising a training sequence comprising a sequence of tokens; processing, by the computing system, at least a portion of the sequence of tokens in the training sequence with a sequence processing model to generate base probabilities respectively for one or more candidate tokens included in a token vocabulary; …” (SHAMIR [0007]).
via the machine learning model, performing probabilistic token-wise generation of other output tokens in the output sequence based on the remainder of the input tokens without the one or more deterministic input tokens;
“The sequence processing model 106 and the base prediction layer 116 generate base probabilities 120, which are the probabilities associated with each candidate token from a predetermined token vocabulary succeeding the input portion of the training sequence 102. These base probabilities 120 are a reflection of the model's assessment of the likelihood of each token's occurrence in the natural progression of the training sequence.” (SHAMIR [0091]).
“Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance quality where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution and pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem. ” (SHAMIR [0234]).
“Prediction layer(s) 6 can evaluate associations between portions of input sequence 5 and a particular output element. These associations can inform a prediction of the likelihood that a particular output follows the input context. For example, consider the textual snippet, “The carpenter's toolbox was small and heavy. It was full of ______.” Example prediction layer(s) 6 can identify that “It” refers back to “toolbox” by determining a relationship between the respective embeddings. Example prediction layer(s) 6 can also link “It” to the attributes of the toolbox, such as “small” and “heavy.” Based on these associations, prediction layer(s) 6 can, for instance, assign a higher probability to the word “nails” than to the word “sawdust.” ” (SHAMIR [0203]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified the invention of _ taught by JSONFORMER with the invention of _ taught by SHAMIR to include further details and narrow down the generation of an output by the machine learning model. The motivation to do so is teach by SHAMIR in “[0234] ...In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.” (SHAMIR [0234]).
Regarding claim 13, the rejection of claim 12 is incorporated, furthermore arguments analogous to claim 2 are applicable.
Regarding claim 14, the rejection of claim 12 is incorporated, furthermore arguments analogous to claim 3 are applicable.
Regarding claim 15, the rejection of claim 14 is incorporated, furthermore arguments analogous to claim 4 are applicable.
Regarding claim 16, the rejection of claim 15 is incorporated, furthermore arguments analogous to claim 5 are applicable.
Regarding claim 17, the rejection of claim 12 is incorporated, furthermore arguments analogous to claim 6 are applicable.
Regarding claim 18, the rejection of claim 12 is incorporated, furthermore arguments analogous to claim 8 are applicable.
Regarding claim 19, the rejection of claim 12 is incorporated, furthermore arguments analogous to claim 9 are applicable.
Regarding claim 20, JSONFORMER teaches:
instantiate a trained machine learning model, configured to: receive an input sequence of input tokens of a prompt, the prompt including context and a structured output definition;
See Fig. 1, JSONFORMER teaches instantiating a trained model, and receiving an input prompt, as well as the prompt including context and output definition in the following lines of code:
“model = AutoModelForCausalLM.from_pretrained("databricks/dolly-v2-12b") “ (JSONFORMER, Fig. 1)
“prompt = "Generate a person's information based on the following schema:" “ (JSONFORMER, Fig. 1)
“jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()” (JSONFORMER, Fig. 1)
identify one or more deterministic input tokens corresponding to the structured output definition;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
in response to identifying the one or more deterministic input tokens: skip transmission of the one or more deterministic input tokens to an input of a transformer of the machine learning model;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
and write the one or more deterministic input tokens as deterministic output tokens to an output token sequence;
“Jsonformer is a new approach to this problem. In structured data, many tokens are fixed and predictable. Jsonformer is a wrapper around HuggingFace models that fills in the fixed tokens during the generation process, and only delegates the generation of content tokens to the language model. This makes it more efficient and bulletproof than existing approaches.” (JSONFORMER, README – “Problem – Solution” section).
“Efficiency: By generating only the content tokens and filling in the fixed tokens, Jsonformer is more efficient than generating a full JSON string and parsing it.” (JSONFORMER, README – “Features” section).
and output the output sequence including the deterministic output tokens and the other output tokens generated by the probabilistic token-wise generation.
“prompt = "Generate a person's information based on the following schema:"
jsonformer = Jsonformer(model, tokenizer, json_schema, prompt)
generated_data = jsonformer()
print(generated_data)” (JSONFORMER, Fig. 1).
JSONFORMER does not explicitly disclose, but SHAMIR teaches:
A computing system, comprising: processing circuitry and associated memory storing instructions that when executed cause the processing circuitry to:
“One example aspect of the present disclosure is directed to a computer-implemented method for performing preference optimization, the method comprising: obtaining, by a computing system comprising one or more computing devices, a training tuple comprising a training sequence comprising a sequence of tokens; processing, by the computing system, at least a portion of the sequence of tokens in the training sequence with a sequence processing model to generate base probabilities respectively for one or more candidate tokens included in a token vocabulary; …” (SHAMIR [0007]).
perform probabilistic token-wise generation of other output tokens in the output sequence based on a remainder of the input tokens without the one or more deterministic input tokens;
“The sequence processing model 106 and the base prediction layer 116 generate base probabilities 120, which are the probabilities associated with each candidate token from a predetermined token vocabulary succeeding the input portion of the training sequence 102. These base probabilities 120 are a reflection of the model's assessment of the likelihood of each token's occurrence in the natural progression of the training sequence.” (SHAMIR [0091]).
“Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance quality where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution and pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem. ” (SHAMIR [0234]).
“Prediction layer(s) 6 can evaluate associations between portions of input sequence 5 and a particular output element. These associations can inform a prediction of the likelihood that a particular output follows the input context. For example, consider the textual snippet, “The carpenter's toolbox was small and heavy. It was full of ______.” Example prediction layer(s) 6 can identify that “It” refers back to “toolbox” by determining a relationship between the respective embeddings. Example prediction layer(s) 6 can also link “It” to the attributes of the toolbox, such as “small” and “heavy.” Based on these associations, prediction layer(s) 6 can, for instance, assign a higher probability to the word “nails” than to the word “sawdust.” ” (SHAMIR [0203]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified the invention of _ taught by JSONFORMER with the invention of _ taught by SHAMIR to include further details and narrow down the generation of an output by the machine learning model. The motivation to do so is teach by SHAMIR in “[0234] ...In this manner, tool use can allow some example models to focus on the strengths of machine-learned models—e.g., understanding an intent in an unstructured request for a task—while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.” (SHAMIR [0234]).
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over JSONFORMER in view of SHAMIR in further view of Henderson; Charles W. et al. (US 20180150752 A1) hereinafter HENDERSON.
Regarding claim 10, the rejection of claim 1 is incorporated, furthermore JSONFORMER teaches:
and display deterministically generated text based on the one or more deterministic output tokens in a visually distinguishable manner from probabilistically generated text based on the other tokens using the deterministic token metadata and/or probabilistic token metadata.
See Fig. 2 and Fig. 3 for identifying the tokens in a visually distinguishable manner, in this case by utilizing different colors for the text generated by a language model and for the text that comes from the deterministic input format.
JSONFORMER in view of SHAMIR does not explicitly disclose, but CHEN teaches:
The computing system of claim 1, wherein the processing circuity is configured to: output deterministic token metadata labeling the one or more deterministic output tokens in the output sequence and/or probabilistic token metadata labeling the other tokens in the output sequence generated by the machine learning model;
“Technologies are described herein for identifying artificial intelligence content. According to some examples, content received is analyzed to determine if the content was generated by an artificial intelligence source. Upon identifying that the content was generated by an artificial intelligence source, the content is displayed in a manner that indicates the determination.” (HENDERSON [Abstract]).
“Returning to the example of analyzing for AI content, if the content analyzer 118 determines that the content received and analyzed is not content provided by an artificial intelligence source, the content analyzer 120 transmits a “no artificial intelligence detected” output to the content identification service 114. The content identification service 114 thereafter causes or allows the content to be provided to the content viewer 106 in the user device 102. If the content analyzer 118 determines that the content received and analyzed is in whole or in part content provided by an artificial intelligence source, the content analyzer 120 transmits an “artificial intelligence detected” output to the content identification service 114. In some examples, the content analyzer 118 can also identify the particular portion of the content containing the artificial intelligence content. The content identification service 114, after receiving the “artificial intelligence detected” can either cause the content from being received by the content viewer 106, thus acting as a filtering service, or provide an output that changes the appearance of the portion of the content identified as being generated from an artificial intelligence source. ” (HENDERSON [0033]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modify taught in JSONFORMER in view of SHAMIR with the invention thought by HENDERSON. The motivation to do so would be to have labels to identify deterministic and generative output text . As suggested by HENDERSON in: ”Technologies are described herein for identifying artificial intelligence content. Generally described, an artificial intelligence (“AI”) content identification system receives content from one or more content providers. Content can be information such as text from web pages, video, documents, speech or aural input, and the like. The AI content identification system analyzes at least a portion of the content received from the one or more content providers to determine if the portion of the content is content generated by an artificial intelligence source. ” (HENDERSON [0033]).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over JSONFORMER in view of SHAMIR in further view of CHEN; Mark et al. (US 20240020096 A1) hereinafter CHEN.
Regarding claim 11, the rejection of claim 1 is incorporated, furthermore JSONFORMER in view of SHAMIR does not explicitly disclose, but CHEN teaches:
The computing system of claim 1, wherein the machine learning model is a transformer-based model including an encoder-decoder architecture, decoder-only architecture, or encoder-only architecture.
“Consistent with some disclosed embodiments, the trained machine learning model may comprise a plurality of layers, at least one of the layers having a transformer decoder architecture. A transformer decoder architecture, as used herein, may refer to a machine learning architecture comprising an encoder and a decoder, both of which are built on a self-attention mechanism, wherein the encoder focuses on encoding the input sequence, and the decoder generates the output sequence based on the encoded representation. ” (CHEN [0071]).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modify taught in JSONFORMER in view of SHAMIR with the invention thought by CHEN. The motivation to do so would be to further limit and define the architecture of the machine learning model and the layers that form it. As suggested by CHEN in: “In some embodiments, the trained machine learning model may comprise a plurality of layers, at least one of the layers having a transformer decoder architecture. ” (CHEN [0015]).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HECTOR J. CRESPO FEBLES whose telephone number is (571)272-4512. The examiner can normally be reached Mon - Fri 7:30 - 5:00.
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, Daniel Washburn can be reached at (571) 272-5551. 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.
/HECTOR J. CRESPO FEBLES/Examiner, Art Unit 2657
/SAMUEL G NEWAY/Primary Examiner, Art Unit 2657
Prosecution Insights