Enforcing self-consistency in language model outputs

Abstract

A device may include one or more memories configured to store input data and one or more processors coupled to the one or more memories and configured to: process, using the machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.

Description

Qualcomm Ref. No. 2500985WO1ENFORCING SELF-CONSISTENCY IN LANGUAGE MODEL OUTPUTSTECHNICAL FIELD

[0001] The present disclosure generally relates to language models. For example, aspects of the present disclosure relate to verifying and ensuring consistency between outputs of language models.BACKGROUND

[0002] Language models may be used to generate text, for example, in response to a user prompt. But some language models may output inconsistent responses, which reduce the utility of and reduce user confidence in language models.SUMMARY

[0003] The following presents a simplified summary' relating to one or more aspects disclosed herein. Thus, the following summary should not be considered an extensive overview relating to all contemplated aspects, nor should the following summary be considered to identify key or critical elements relating to all contemplated aspects or to delineate the scope associated with any particular aspect. Accordingly, the following summary presents certain concepts relating to one or more aspects relating to the mechanisms disclosed herein in a simplified form to precede the detailed description presented below.

[0004] In some aspects, an apparatus for detecting inconsistencies between outputs of a machine learning model is provided. The apparatus includes one or more memories configured to store input data and one or more processors coupled to the one or more memories and configured to: process, using the machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO2

[0005] In some aspects, a method for detecting inconsistencies between outputs of a machine learning model is provided. The method includes: processing, using a machine learning model, a first input to generate a first output; processing, using the machine learning model, a second input to generate a second output; determining a contradiction score representing a contradiction between the first output and the second output; determining that a contradiction exists between the first output and the second output based on the contradiction score; and adjusting the first output based on a determination the contradiction exists between the first output and the second output.

[0006] In some aspects, a non-transitoiy computer-readable medium is provided having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to: process, using a machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.

[0007] In another example, an apparatus for detecting inconsistencies between outputs of a machine learning model is provided. The apparatus includes: means for processing, using a machine learning model, a first input to generate a first output; means for processing, using the machine learning model, a second input to generate a second output; means for determining a contradiction score representing a contradiction between the first output and the second output; means for determining that a contradiction exists between the first output and the second output based on the contradiction score; and means for adjusting the first output based on a determination the contradiction exists between the first output and the second output.

[0008] In some aspects, one or more of the apparatuses described herein is, can be part of, or can include an extended reality device (e.g., a virtual reality (VR) device, an augmented reality (AR) device, or a mixed reality (MR) device), a vehicle (or a computing device, system, or component of a vehicle), a mobile device (e.g., a mobile telephone or so-called “smart phone”, a tablet computer, or other type of mobile device), a smart orPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO3 connected device (e.g., an Intemet-of-Things (loT) device), a wearable device, a personal computer, a laptop computer, a video server, a television (e.g., a network-connected television), a robotics device or system, or other device. In some aspects, each apparatus can include an image sensor (e.g., a camera) or multiple image sensors (e.g.. multiple cameras) for capturing one or more images. In some aspects, each apparatus can include one or more displays for displaying one or more images, notifications, and / or other displayable data. In some aspects, each apparatus can include one or more speakers, one or more light-emitting devices, and / or one or more microphones. In some aspects, each apparatus can include one or more sensors. In some cases, the one or more sensors can be used for determining a location of the apparatuses, a state of the apparatuses (e.g., a tracking state, an operating state, a temperature, a humidity level, and / or other state), and / or for other purposes.

[0009] This summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used in isolation to determine the scope of the claimed subject matter. The subject matter should be understood by reference to appropriate portions of the entire specification of this patent, any or all drawings, and each claim.

[0010] The foregoing, together with other features and aspects, will become more apparent upon referring to the following specification, claims, and accompanying drawings.BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Illustrative examples of the present application are described in detail below with reference to the following figures:

[0012] FIG. 1 is a block diagram illustrating an example of a system through which a user may interact, according to various aspects of the present disclosure;

[0013] FIG. 2 is a block diagram illustrating an example of a system using a language model, according to various aspects of the present disclosure;

[0014] FIG. 3 is a block diagram illustrating an example of a system for adjusting responses of a language model, according to various aspects of the present disclosure;Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO4

[0015] FIG. 4 is a block diagram illustrating an example of a system for detecting similarity and contradiction between responses of a language model, according to various aspects of the present disclosure;

[0016] FIG. 5 is a block diagram illustrating an example of a system for adjusting responses of a language model, according to various aspects of the present disclosure;

[0017] FIG. 6 is a block diagram illustrating an example of a system for adjusting responses of a language model, according to various aspects of the present disclosure;

[0018] FIG. 7 is a flow diagram illustrating an example of a process, in accordance with aspects of the present disclosure:

[0019] FIG. 8 is a block diagram illustrating an example of a deep learning neural network that can be used to perform various tasks, according to some aspects of the disclosed technology;

[0020] FIG. 9 is a block diagram illustrating an example of a convolutional neural network (CNN), according to various aspects of the present disclosure;

[0021] FIG. 10 is a block diagram of an example transformer in accordance with some aspects of the disclosure; and

[0022] FIG. 1 1 is a block diagram illustrating an example computing-device architecture of an example computing device which can implement the various techniques described herein.DETAILED DESCRIPTION

[0023] Certain aspects of this disclosure are provided below. Some of these aspects may be applied independently and some of them may be applied in combination as would be apparent to those of skill in the art. In the following description, for the purposes of explanation, specific details are set forth to provide a thorough understanding of aspects of the application. However, it will be apparent that various aspects may be practiced without these specific details. The figures and description are not intended to be restrictive.

[0024] The description provides example aspects only, and is not intended to limit the scope, applicability, or configuration of the disclosure. Rather, the description of thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO5 exemplary aspects will provide those skilled in the art with an enabling description for implementing an exemplary aspect. Various changes may be made in the function and arrangement of elements without departing from the spirit and scope of the application as set forth in the appended claims.

[0025] The terms “exemplary"’ and / or “example” are used herein to mean “serving as an example, instance, or illustration.” Any aspect described herein as "‘exemplary” and / or “example” is not necessarily to be construed as preferred or advantageous over other aspects. Likewise, the term “aspects of the disclosure” does not require that all aspects of the disclosure include the discussed feature, advantage, or mode of operation.

[0026] Systems, apparatuses, methods (also referred to as processes), and computer- readable media (collectively referred to herein as “systems and techniques”) are described herein for detecting and resolving contradictions between responses of machine learning models, such as language models. One example of a language model is a Large Language Models (LLMs). Language models may be used to generate text, for example, in response to a user query and / or analysis of an audio input. However, language models may occasionally output inconsistent responses. Such behavior may not meet user expectations and may reduce confidence in a language model-based system.

[0027] The systems and techniques described herein can identify and correct inconsistent responses generated by machine learning models, including language models. For instance, a system can measure a contradiction score for one or more pairs of responses received from a language model. The contradiction score indicates whether the pairs of responses contain a contradiction. If a contradiction is identified, then the system can adjust one or more of the responses accordingly.

[0028] Various aspects of the application will be described with respect to the figures below.

[0029] FIG. 1 is a block diagram illustrating an example of a system 100 through which a user may interact, according to various aspects of the present disclosure. In the example depicted, the system 102 receives audio data 104 and / or a user input 106. In turn, the system 102 generates one or more responses 108. As discussed further herein, the systemPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO6102 can include one or more machine learning models such as language models or large language models.

[0030] In an example, the user input 106 may take the form of one or more queries or questions, which are in turn answered by the responses 108. The user may provide user input 106 to system 102 through any suitable format, for example, the user may interact with system 102 via a text chat system or through an audio chat system. For example, the user may enter user input 106 as a text query into a text-based chat application. As another example, the user may speak user input 106 and a microphone may record user input 106 and convert the spoken user input into a text format. User input 106 may relate to audio events of audio data 104. For example, user input 106 may include a user input regarding the audio events, such as “does audio data 104 include a dog barking?” or “at what time of audio data 104 does a bell ring?”

[0031] System 102 may generate one or more responses 108 based on user input 106 and responsive to user input 106. Response 108 may have any suitable format. In some aspects, response 108 may have the same format as user input 106. For example, response 108 may be a text response displayed at a display of a text-based chat application. As another example, response 108 may be a vocalized response output by a speaker.

[0032] As further explained, the system 102 may process audio data 104, which may be referred to by the user input 106 and the analysis of which may play a part in the responses. Audio data 104 may include audio events. For example, audio data 104 may include sound recordings of various events, such as a person speaking, a dog barking, a bell ringing, etc. Audio data 104 may be according to any suitable format, such as Motion- Picture Experts Group (MPEG) Audio Layer III (MP3), waveform audio file format (WAV), and so forth.

[0033] Audio data 104 may be labeled, or tagged, with metadata by system 102 or by another system. For example, system 102 may generate audio metadata from audio data 104 and attach the metadata to the audio. In some cases, labeling may be accomplished via one or more a machine-learning models trained to label audio data, such as an audio classification device (ACD). For example, labeling may be performed by a classifier network trained to classify audio data into various pre-determined classes.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO7

[0034] The audio data may include, for each audio event, an identifier (e.g., a numerical identifier (ID)), a label, a start time, a stop time, a duration, and / or an occurrence indicator. In some aspects, the audio metadata may be generated, stored, and / or updated according to a JavaScript Object Notation (JSON) format. The label of an audio event may be based on the class into the audio event was classified. The label of an audio event may be descriptive of the audio event. For example, an audio event may be labeled “silence,” “church bell,” “bird chirping,” etc.

[0035] The start time of an audio event may indicate a time, relative to the audio data 104 (e.g., t = 2.4 seconds after the beginning of audio data 104), or relative to an absolute date and time, of the beginning of the audio event. The stop time of an audio event may indicate a time of the end of the audio event. The duration of an audio event may be a time difference between when the audio event began and when the audio event ceased. An audio event may occur multiple times in audio data. For example, a single audio file may include recordings of three separate instances of car horns honking. The occurrence indicator of the audio metadata may indicate which occurrence of an audio event an entry in the audio metadata refers to. For example, for an instance of audio data, a first entry of audio metadata may have an ID of “1,” a label of “dog barking,” a start time of “5.0” (e.g., t = 5 seconds after the beginning of the instance of audio data 104), and an occurrence indicator of “1” (e.g., indicating this is the first entry in audio metadata related to the label “dog barking”). A second entry of audio metadata may have an ID of “2,” a label of “dog barking,” a start time of “35.0,” and an occurrence indicator of “2” (e g., indicating that this is the second entry in the audio metadata related to the label “dog barking”).

[0036] As discussed, certain aspects detect and resolve contradictions between responses of language models. The responses can be based on audio analyzed by the language model. The examples below illustrate inconsistent responses from a language model. For instance, in the first example below, the system indicates that the audio does not contain sound of a chain rattling, but the audio does in fact include such a sound. When corrected by the user, the model contradicts itself.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO8

[0037] In the second example below, the system answers that the recording is not a live recording. But when asked again later, the system states that the recording is a live recording.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO9

[0038] In the third example below, the system fails to state that the music includes an alarm bell.

[0039] In some cases, the system 102 can adjust or correct a response generated by a machine learning model, thereby improving the accuracy of the system. Such an example is discussed further with respect to FIGs. 2-6.

[0040] FIG. 2 is a block diagram illustrating an example of a system 200 using a language model, according to various aspects of the present disclosure. In the example depicted, language model 214 receives one or more user inputs, or questions 202 and inPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO10 turn generates one or more responses 204. which as explained below, may be adjusted by response adjuster 216.

[0041] Language model 214 is configured to receive queries (questions) 202 (as depicted Qi... Qi) and generate one or more responses 204 (Ri... Ri). Language model 214 may be trained to generate responses based on inputs, context data, and / or prompts (e.g., text prompts). An example of a language model 214 is a large-language model (LLM). Language model 214 may generate responses in response to a query from a user device. In some cases, language model 214 may generate responses based on text identified from an audio input.

[0042] In some aspects, system 200 may store, maintain, and / or update a history of inputs based on user interactions with system 200. For example, a history can include the questions Qi... Qi and / or the responses Ri... Ri. The history may be used by response adjuster 216, for example, to determine consistency between responses.

[0043] In an example, response adjuster 216 processes the responses (Ri... Ri) to verify consistency between one or more pairs of the responses (e.g., Ri and R2, and so forth). If an inconsistency is identified then the response adjusted can cause the responses to be adjusted accordingly, for example, by prompting the language model 214 to adjust the responses. Responses can be corrected one at a time, or in batch form.

[0044] In some cases, system 200 may generate a prompt, which may be provided to language model 214 to adjust the behavior of language model 214 and / or refine the generated responses. For example, a prompt may include instructions for language model 214 to revise one or more of the responses. The prompt may include instructions for language model 214 to use more or less of the history as additional contextual information for responding to user questions 202.

[0045] FIG. 3 is a block diagram illustrating an example of a system 300 for adjusting responses of a language model, according to various aspects of the present disclosure. Relative to system 300, system 300 illustrates contradiction checkers 310a-n, each of which can verify a presence of a contradiction in pairs of sentences or parts thereof.

[0046] In the example depicted, system 300 receives responses 304 (e.g., Ri . . . Ri) and verifies various pairs of responses using contradiction checkers 310a-n. If thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO11 contradiction checkers 310a-n identify any pairs of inconsistent responses, then one or both of the responses of the pair are provided to response adjuster 316.

[0047] For example, contradiction checker 310a receives two inputs, response Ri and Ri, where R, is the last responses in a group of responses of length i. Similarly, contradiction checker 310b receives R2 and Ri, and contradiction checker 310n receives response Ri and R . Each contradiction checker 310a-n outputs a respective indication of whether contradiction is detected. While three contradiction checkers are depicted, any number of responses may be checked. In some case, fewer than a total of responses in a group may be analyzed, for example, to save computational resources. If no contradictions are identified, then no responses are provided to the response adjuster 316.

[0048] In some cases, system 300 may check a subset of available sentences, to save on computational usage. For instance, instead of comparing every response Ri with all past responses, system 300 may only compare Ri with the last few responses. In another example, system 300 can verify pairs of responses on a cadence, for example, every few responses.

[0049] Different approaches may be used to identify contradiction between a pair of responses. In some cases, contradiction checkers 3f 0a-n measure similarity in addition to contradiction. In this manner, contradiction checkers 310a-n may verify whether two sentences or fragments thereof have sufficient similarity, indicating that the sentences identify a similar or identical topic. FIG. 4 depicts an example of a contradiction checker.

[0050] FIG. 4 is a block diagram illustrating an example of a system 400 for detecting similarity and contradiction between responses of a language model, according to various aspects of the present disclosure. In the example depicted in FIG. 4, similarity calculator 410 and contradiction calculator 420 are used to identify a presence of a contradiction between input 402 and input 404. System 400 further includes a first threshold checker 414 that checks similarity score 412 and a second threshold checker 416 that checks contradiction score 422.

[0051] Different approaches to determining similarity and contradiction are possible. For instance, similarity and / or contradiction may be determined from entire sentences orPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO12 fragments thereof. Therefore, the inputs 402 and 404 may be segments, fragments of sentences, phrases, or one or more words.

[0052] Similarity calculator 410 receives inputs 402 and 404 and determines similarity score 412. Similarity score 412 can represent a similarity between inputs, for example, reflecting an overlap in content between the sentences or fragments. Similarity7score 412 can be determined by forming a text embedding for each input 402 and 404. In some cases, averages are computed for text embedding. An embedding is a vector that represents a sentence. The similarity score 412 can be derived from the averages.

[0053] Other similarity metrics that may be used include Bilingual Evaluation Understudy (BLEU), Recall-Oriented Understudy for Gisting Evaluation (ROUGE), Metric for Evaluation of Translation with Explicit Ordering (METEOR), Consensusbased Image Description Evaluation (CIDER), Semantic Propositional Image Caption Evaluation (SPICE), SP1CE++, and an n-gram overlap An additional machine learning model or language model can also be used to measure content overlap / similarity.

[0054] Similarity score 412 may be compared against a threshold by first threshold checker 414. If similarity score 412 is beyond the threshold, then a similarity exists between the inputs 402 and 404. A measure of similarity in conjunction with a measure of contradiction is helpful because without similarity, a true contradiction cannot be detected.

[0055] Contradiction calculator 420 receives inputs 402 and 404 and determines contradiction score 422. Contradiction score 422 represents a level of contradiction between inputs 402 and 404. Different approaches to determining contradiction score 422 are possible. For instance, contradiction calculator 420 may use machine learning approaches. A model may be trained to predict contradiction, entailment, and / or neutrality between sentences or fragments thereof. The model may be trained with various text embeddings. An example of neutrality is Recognizing Textual Entailment (RTE). The RTE model has pre-trained text embedding layers followed by some layers fine-tuned for reaching this judgement. A language model can also be used to detect / measure contradiction. Natural language inference can be used.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO13

[0056] Contradiction score 422 may be compared against a threshold by second threshold checker 416 . If contradiction score 422 is beyond the threshold, then a contradiction exists between the inputs 402 and 404. System 400 can determine whether both the similarity score 412 and contradiction score 422 are beyond their respective thresholds, then system 400 identifies a contradiction between input 402 and input 404. In some cases, a high entailment score may imply a low contradiction score, and vice versa. An entailment score represents a degree to which one statement can be logically inferred from another statement.

[0057] If either similarity score 412 and / or 422 are below the respective thresholds, then the process ends at 420. If both similarity score 412 and 422 are above the respective thresholds as verified by “AND” operator 418, then the process continues.

[0058] In some cases, a user can configure one or more of the parameters of machine learning models used to determine similarity and / or contradiction. For instance, a user may provide feedback that a particular pair of sentences were contradictory when the system did not identify the sentences as such, and vice versa. This can cause the internal parameters of the model to be updated, thereby resulting in better predictions.

[0059] FIG. 5 is a block diagram illustrating an example of a system 500 for adjusting responses of a language model, according to various aspects of the present disclosure. In the example depicted, system 500 adjusts one or more responses that have been identified as inconsistent.

[0060] System 500 includes response adjusters 516a-b, which can communicate with a language model such as language model 214. The response adjusters 516a-b can use one or more prompts 506 to instruct the language model to adjust the responses. For instance, a prompt can instruct the model to adjust specific responses.

[0061] Response adjuster 516a receives responses Ri . .. Rk, where k represents a number of responses to be adjusted. The value k may be less than the value i. which represents the total responses. Continuing the example, response adjuster 516a adj usts responses Ri . .. Rk as appropriate. For instance, response adjuster 516a can determine, between one or more pairs of responses, whether an inconsistency exists.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO14

[0062] Response adjuster 516a assumes that if a contradiction exists, then the first response in the pair of responses (e.g., R2 in a pair of responses Ri and R2) is correct. But other examples are possible. For example, the system may assume that a second response in a pair of responses is correct whereas the first response is incorrect. System 500 may further include an additional optional response adjuster 516b, which can be used to verify the corrected responses again. System 500 outputs revised responses 518.

[0063] FIG. 6 is a block diagram illustrating an example of a system 600 for adjusting responses of a language model, according to various aspects of the present disclosure. In the example depicted, system 600 adjusts one or more responses 601 that have been identified as inconsistent based on feedback from a user device.

[0064] In some aspects, system 600 can receive indications and / or directions from a user device. For instance, system 602 can transmit a prompt (e.g., a message) to a user device requesting that a user identify' one or more responses that are inconsistent or incorrect. In the example depicted, responses denoted as Rincorrect a and Rincon ect b are identified as incorrect based on feedback (e.g., as indicated by a user based on user input provided via an input interface of the user device). Continuing the example, response adjuster 616a acts on this feedback. Response adjuster 616a provides prompt 606 to the language model. The prompt 606 requests the language model to revise its response. The language model in turn outputs revised responses denoted as Rincorrect ajrevised) and Rincorrect b(revised). This process can continue, for example, the revised responses may be optionally provided to response adjuster 616b, which can verify consistency. Other examples are possible. System 500 outputs revised responses 618.

[0065] FIG. 7 is a flow diagram illustrating an example of a process 700, in accordance with aspects of the present disclosure. One or more operations of process 700 may be performed by a computing device (or apparatus) or a component (e.g., one or more processors, a chipset, a system-on-chip (SOC), codec, etc.) of the computing device. The computing device may be a mobile device (e.g., a mobile phone), a network-connected wearable such as a watch, an extended reality (XR) device such as a virtual reality (VR) device or augmented reality (AR) device, a vehicle or component or system of a vehicle, a desktop computing device, a tablet computing device, a server computer, a robotic device, and / or any other computing device with the resource capabilities to perform thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO15 one or more operations of process 700. The one or more operations of process 700 may be implemented as software components that are executed and run on one or more processors.

[0066] At block 702, the computing device (or component thereof) can process, using the machine learning model, a first input to generate a first output. In some aspects, the first input includes a first text embedding associated with a first question. In some aspects, the first output includes text associated with a first answer to the first question.

[0067] At block 704, the computing device (or component thereof) can process, using the machine learning model, a second input to generate a second output. In some aspects, the second input includes a second text embedding associated with a second question. In some aspects, the second output includes text associated with a second answer to the second question.

[0068] At block 706, the computing device (or component thereof) can determine a contradiction score representing a contradiction between the first output and the second output. In some aspects, determining the contradiction score can involve compare the first output and the second output to determine at least one of an entailment value, a contradiction value, or a neutrality value; and determining the contradiction score as the contradiction value.

[0069] At block 708, the computing device (or component thereof) can determine that a contradiction exists between the first output and the second output based on the contradiction score.

[0070] At block 710, the computing device (or component thereof) can adjust the first output based on a determination the contradiction exists between the first output and the second output. In some aspects, provide the adjusted first output and the second output to a user device. In some aspects, the system can configure a display to display the adjusted first output and / or the second output.

[0071] In some aspects, to adjust the output, the computing device (or component thereof) can process, using the machine learning model, the first output and a prompt to generate the adjusted first output, wherein the prompt includes instructions for the machine learning model to revise the first output.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO16

[0072] In some aspects, the computing device (or component thereof) can determine that the contradiction exists further based on a user input indicative of the contradiction received from a user device. In some aspects, to determine that the contradiction exists includes the computing device (or component thereof) can process, using an additional machine learning model, the first output and the second output to generate an indication of a contradiction between the first output and the second output, wherein the additional machine learning model is trained to detect textual similarities.

[0073] In some cases, a similarity score is used. The similarity score can represent a similarity7between the first output and the second output. The similarity' score can be determined via various approaches. For instance, in some aspects, the computing device (or component thereof) can determine a similarity score representing a similarity between the first output and the second output. The computing device (or component thereof) can use the similarity' score to determine that the contradiction exists.

[0074] In some aspects, to determine the similarity score, the computing device (or component thereof) can determine a first text embedding of the first output and a second text embedding of the second output. The computing device (or component thereof) can then determine the similarity score based on a comparison between the first text embedding and the second text embedding.

[0075] In some aspects, to determine the score, the computing device (or component thereof) can process the first output and the second output using an additional machine learning model. The additional machine learning model can be trained to detect textual similarities.

[0076] In an aspect, to determine a similarity score, the computing device (or component thereof) can compare the similarity score to a similarity’ threshold and the contradiction score to a contradiction threshold. The computing device (or component thereof) can determine that the contradiction exists based on a determination that the similarity' score is greater than the similarity’ threshold and a determination that the contradiction score is greater than the contradiction threshold.

[0077] In some cases, the machine learning model may be reset. For instance, the computing device (or component thereof) can determine one or more additionalPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO17 contradictions between at least two outputs, determine that a number of the one or more additional contradictions is greater than an additional threshold, and reset the machine learning model based on the number of the one or more additional contradictions being greater than the additional threshold.

[0078] In some examples, as noted previously, the methods described herein (e.g., process 700 of FIG. 7, and / or other methods described herein) can be performed, in whole or in part, by a computing device or apparatus. In one example, one or more of the methods can be performed by systems depicted in FIGs. 1-6, or by another system or device. In another example, one or more of the methods (e.g., process 700, and / or other methods described herein) can be performed, in whole or in part, by the computing-device architecture 1100 shown in FIG. 11. For instance, a computing device with the computing-device architecture 1100 shown in FIG. 11 can include, or be included in, the components of the systems described herein and can implement the operations of process 700, and / or other process described herein. In some cases, the computing device or apparatus can include various components, such as one or more input devices, one or more output devices, one or more processors, one or more microprocessors, one or more microcomputers, one or more cameras, one or more sensors, and / or other component(s) that are configured to carry' out the steps of processes described herein. In some examples, the computing device can include a display, a network interface configured to communicate and / or receive the data, any’ combination thereof, and / or other component(s). The network interface can be configured to communicate and / or receive Internet Protocol (IP) based data or other type of data.

[0079] The components of the computing device can be implemented in circuitry. For example, the components can include and / or can be implemented using electronic circuits or other electronic hardware, which can include one or more programmable electronic circuits (e.g., microprocessors, graphics processing units (GPUs), digital signal processors (DSPs), central processing units (CPUs), and / or other suitable electronic circuits), and / or can include and / or be implemented using computer software, firmware, or any combination thereof, to perform the various operations described herein.

[0080] Process 700, and / or other process described herein are illustrated as logical flow diagrams, the operation of which represents a sequence of operations that can bePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO18 implemented in hardware, computer instructions, or a combination thereof. In the context of computer instructions, the operations represent computer-executable instructions stored on one or more computer-readable storage media that, when executed by one or more processors, perform the recited operations. Generally, computer-executable instructions include routines, programs, objects, components, data structures, and the like that perform particular functions or implement particular data types. The order in which the operations are described is not intended to be construed as a limitation, and any number of the described operations can be combined in any order and / or in parallel to implement the processes.

[0081] Additionally, process 700, and / or other process described herein can be performed under the control of one or more computer systems configured with executable instructions and can be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) executing collectively on one or more processors, by hardware, or combinations thereof. As noted above, the code can be stored on a computer-readable or machine-readable storage medium, for example, in the form of a computer program comprising a plurality of instructions executable by one or more processors. The computer-readable or machine-readable storage medium can be non- transitory.

[0082] As noted above, various aspects of the present disclosure can use machinelearning models or systems.

[0083] FIG. 8 is an illustrative example of a neural network 800 (e.g., a deep-learning neural network) that can be used to implement machine-learning based feature segmentation, implicit-neural -representation generation, rendering, classification, object detection, image recognition (e.g., face recognition, object recognition, scene recognition, etc.), feature extraction, authentication, gaze detection, gaze prediction, and / or automation. For example, neural network 800 may be an example of, or can implement functionality in systems depicted in FIGs. 1-6.

[0084] An input layer 802 includes input data. In one illustrative example, input layer 802 can include data representing audio data 104 of FIG. 1. Neural network 800 includes multiple hidden layers, for example, hidden layers 806a, 806b, through 806n. The hidden layers 806a, 806b, through hidden layer 806n include ‘n” number of hidden layers, wherePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO19“n” is an integer greater than or equal to one. The number of hidden layers can be made to include as many layers as needed for the given application. Neural network 800 further includes an output layer 804 that provides an output resulting from the processing performed by the hidden layers 806a, 806b, through 806n. In one illustrative example, output layer 804 can generate audio metadata for audio 104.

[0085] Neural network 800 may be, or may include, a multi-layer neural network of interconnected nodes. Each node can represent a piece of information. Information associated with the nodes is shared among the different layers and each layer retains information as information is processed. In some cases, neural network 800 can include a feed-forward network, in which case there are no feedback connections where outputs of the network are fed back into itself. In some cases, neural network 800 can include a recunent neural network, which can have loops that allow information to be carried across nodes while reading in input.

[0086] Information can be exchanged between nodes through node-to-node interconnections between the various layers. Nodes of input layer 802 can activate a set of nodes in the first hidden layer 806a. For example, as shown, each of the input nodes of input layer 802 is connected to each of the nodes of the first hidden layer 806a. The nodes of first hidden layer 806a can transform the information of each input node by applying activation functions to the input node information. The information derived from the transformation can then be passed to and can activate the nodes of the next hidden layer 806b. which can perform their own designated functions. Example functions include convolutional, up-sampling, data transformation, and / or any other suitable functions. The output of the hidden layer 806b can then activate nodes of the next hidden layer, and so on. The output of the last hidden layer 806n can activate one or more nodes of the output layer 804, at which an output is provided. In some cases, while nodes (e.g., node 808) in neural network 800 are shown as having multiple output lines, a node has a single output and all lines shown as being output from a node represent the same output value.

[0087] In some cases, each node or interconnection between nodes can have a weight that is a set of parameters derived from the training of neural network 800. Once neural network 800 is trained, it can be referred to as a trained neural network, which can be used to perform one or more operations. For example, an interconnection between nodesPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO20 can represent a piece of information learned about the interconnected nodes. The interconnection can have a tunable numeric weight that can be tuned (e.g., based on a training dataset), allowing neural network 800 to be adaptive to inputs and able to leam as more and more data is processed.

[0088] Neural network 800 may be pre-trained to process the features from the data in the input layer 802 using the different hidden layers 806a, 806b. through 806n in order to provide the output through the output layer 804. In an example in which neural network 800 is used to identify features in images, neural network 800 can be trained using training data that includes both images and labels, as described above. For instance, training images can be input into the network, with each training image having a label indicating the features in the images (for the feature-segmentation machine-learning system) or a label indicating classes of an activity in each image. In one example using object classification for illustrative purposes, a training image can include an image of a number 2, in which case the label for the image can be [0 0 1 0 0 0 0 0 0 0],

[0089] In some cases, neural network 800 can adjust the weights of the nodes using a training process called backpropagation. As noted above, a backpropagation process can include a forward pass, a loss function, a backward pass, and a weight update. The forward pass, loss function, backward pass, and parameter update are performed for one training iteration. The process can be repeated for a certain number of iterations for each set of training images until neural network 800 is trained well enough so that the weights of the layers are accurately tuned.

[0090] For the example of identifying objects in images, the forward pass can include passing a training image through neural network 800. The weights are initially randomized before neural network 800 is trained. As an illustrative example, an image can include an array of numbers representing the pixels of the image. Each number in the array can include a value from 0 to 288 describing the pixel intensify at that position in the array. In one example, the array can include a 28 x 28 x 3 array of numbers with 28 rows and 28 columns of pixels and 3 color components (such as red, green, and blue, or luma and two chroma components, or the like).

[0091] As noted above, for a first training iteration for neural network 800, the output will likely include values that do not give preference to any particular class due to thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO21 weights being randomly selected at initialization. For example, if the output is a vector with probabilities that the object includes different classes, the probability value for each of the different classes can be equal or at least very similar (e.g., for ten possible classes, each class can have a probability' value of 0.1). With the initial weights, neural network 800 is unable to determine low-level features and thus cannot make an accurate determination of what the classification of the object might be. A loss function can be used to analyze error in the output. Any suitable loss function definition can be used, such as a cross-entropy loss. Another example of a loss function includes the mean squared error (MSE), defined as Etotai = S A (target - output)2. The loss can be set to be equal to the value of Etotai.

[0092] The loss (or error) will be high for the first training images since the actual values will be much different than the predicted output. The goal of training is to minimize the amount of loss so that the predicted output is the same as the training label. Neural network 800 can perform a backward pass by determining which inputs (weights) most contributed to the loss of the network and can adjust the weights so that the loss decreases and is eventually minimized. A derivative of the loss with respect to the weights (denoted as dL / dW, where W are the weights at a particular layer) can be computed to determine the weights that contributed most to the loss of the network. After the derivative is computed, a weight update can be performed by updating all the weights of the filters. For example, the weights can be updated so that they change in the opposite direction of the gradient. The weight update can be denoted as w = Wi - t dL / dW, where w denotes a weight, Wi denotes the initial weight, and denotes a learning rate. The learning rate can be set to any suitable value, with a high learning rate including larger weight updates and a lower value indicating smaller weight updates.

[0093] Neural network 800 can include any suitable deep network. One example includes a convolutional neural network (CNN), which includes an input layer and an output layer, with multiple hidden layers between the input and out layers. The hidden layers of a CNN include a series of convolutional, nonlinear, pooling (for downsampling), and fully connected layers. Neural network 800 can include any other deep netw ork other than a CNN. such as an autoencoder, a deep belief nets (DBNs), a Recurrent Neural Networks (RNNs), among others.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO22

[0094] FIG. 9 is an illustrative example of a convolutional neural network (CNN) 900. The input layer 902 of the CNN 900 includes data representing an image or frame. For example, the data can include an array of numbers representing the pixels of the image, with each number in the array including a value from 0 to 255 describing the pixel intensity at that position in the array. Using the previous example from above, the array can include a 28 x 28 x 3 array of numbers with 28 rows and 28 columns of pixels and 3 color components (e.g., red, green, and blue, or luma and two chroma components, or the like). The image can be passed through a convolutional hidden layer 904, an optional nonlinear activation layer, a pooling hidden layer 906, and fully connected layer 908 (which fully connected layer 908 can be hidden) to get an output at the output layer 910. While only one of each hidden layer is shown in FIG. 9, one of ordinary skill will appreciate that multiple convolutional hidden layers, non-linear layers, pooling hidden layers, and / or fully connected layers can be included in the CNN 900. As previously described, the output can indicate a single class of an object or can include a probability of classes that best describe the object in the image.

[0095] The first layer of the CNN 900 can be the convolutional hidden layer 904. The convolutional hidden layer 904 can analyze image data of the input layer 902. Each node of the convolutional hidden layer 904 is connected to a region of nodes (pixels) of the input image called a receptive field. The convolutional hidden layer 904 can be considered as one or more filters (each filter corresponding to a different activation or feature map), with each convolutional iteration of a filter being a node or neuron of the convolutional hidden layer 904. For example, the region of the input image that a filter covers at each convolutional iteration would be the receptive field for the filter. In one illustrative example, if the input image includes a 28x28 array, and each filter (and corresponding receptive field) is a 5x5 array, then there will be 24x24 nodes in the convolutional hidden layer 904. Each connection between a node and a receptive field for that node learns a weight and, in some cases, an overall bias such that each node learns to analyze its particular local receptive field in the input image. Each node of the convolutional hidden layer 904 will have the same weights and bias (called a shared weight and a shared bias). For example, the filter has an array of weights (numbers) and the same depth as the input. A filter will have a depth of 3 for an image frame examplePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO23(according to three color components of the input image). An illustrative example size of the filter array is 5 x 5 x 3, corresponding to a size of the receptive field of a node.

[0096] The convolutional nature of the convolutional hidden layer 904 is due to each node of the convolutional layer being applied to its corresponding receptive field. For example, a filter of the convolutional hidden layer 904 can begin in the top-left comer of the input image array and can convolve around the input image. As noted above, each convolutional iteration of the filter can be considered a node or neuron of the convolutional hidden layer 904. At each convolutional iteration, the values of the filter are multiplied with a corresponding number of the original pixel values of the image (e.g., the 5x5 filter array is multiplied by a 5x5 array of input pixel values at the top-left comer of the input image array). The multiplications from each convolutional iteration can be summed together to obtain a total sum for that iteration or node. The process is next continued at a next location in the input image according to the receptive field of a next node in the convolutional hidden layer 904. For example, a filter can be moved by a step amount (referred to as a stride) to the next receptive field. The stride can be set to 1 or any other suitable amount. For example, if the stride is set to 1, the filter will be moved to the right by 1 pixel at each convolutional iteration. Processing the filter at each unique location of the input volume produces a number representing the filter results for that location, resulting in a total sum value being determined for each node of the convolutional hidden layer 904.

[0097] The mapping from the input layer to the convolutional hidden layer 904 is referred to as an activation map (or feature map). The activation map includes a value for each node representing the filter results at each location of the input volume. The activation map can include an array that includes the various total sum values resulting from each iteration of the filter on the input volume. For example, the activation map will include a 24 x 24 array if a 5 x 5 filter is applied to each pixel (a stride of 1) of a 28 x 28 input image. The convolutional hidden layer 904 can include several activation maps in order to identify multiple features in an image. The example shown in FIG. 9 includes three activation maps. Using three activation maps, the convolutional hidden layer 904 can detect three different kinds of features, with each feature being detectable across the entire image.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO24

[0098] In some examples, a non-linear hidden layer can be applied after the convolutional hidden layer 904. The non-linear layer can be used to introduce nonlinearity to a system that has been computing linear operations. One illustrative example of a non-linear layer is a rectified linear unit (ReLU) layer. A ReLU layer can apply the function f(x) = max(0, x) to all of the values in the input volume, which changes all the negative activations to 0. The ReLU can thus increase the non-linear properties of the CNN 900 without affecting the receptive fields of the convolutional hidden layer 904.

[0099] The pooling hidden layer 906 can be applied after the convolutional hidden layer 904 (and after the non-linear hidden layer when used). The pooling hidden layer 906 is used to simplify the information in the output from the convolutional hidden layer 904. For example, the pooling hidden layer 906 can take each activation map output from the convolutional hidden layer 904 and generates a condensed activation map (or feature map) using a pooling function. Max-pooling is one example of a function performed by a pooling hidden layer. Other forms of pooling functions be used by the pooling hidden layer 906, such as average pooling, L2-norm pooling, or other suitable pooling functions. A pooling function (e.g., a max-pooling filter, an L2-norm filter, or other suitable pooling filter) is applied to each activation map included in the convolutional hidden layer 904. In the example shown in FIG. 9, three pooling filters are used for the three activation maps in the convolutional hidden layer 904.

[0100] In some examples, max-pooling can be used by applying a max-pooling filter (e.g., having a size of 2x2) with a stride (e g., equal to a dimension of the filter, such as a stride of 2) to an activation map output from the convolutional hidden layer 904. The output from a max-pooling filter includes the maximum number in every sub-region that the filter convolves around. Using a 2x2 filter as an example, each unit in the pooling layer can summarize a region of 2x2 nodes in the previous layer (with each node being a value in the activation map). For example, four values (nodes) in an activation map will be analyzed by a 2x2 max -pooling filter at each iteration of the filter, with the maximum value from the four values being output as the “max” value. If such a max-pooling filter is applied to an activation filter from the convolutional hidden layer 904 having a dimension of 24x24 nodes, the output from the pooling hidden layer 906 will be an array of 12x12 nodes.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO25

[0101] In some examples, an L2-norm pooling filter could also be used. The L2-norm pooling filter includes computing the square root of the sum of the squares of the values in the 2x2 region (or other suitable region) of an activation map (instead of computing the maximum values as is done in max-pooling) and using the computed values as an output.

[0102] The pooling function (e.g., max-pooling, L2-norm pooling, or other pooling function) determines whether a given feature is found anywhere in a region of the image and discards the exact positional information. This can be done without affecting results of the feature detection because, once a feature has been found, the exact location of the feature is not as important as its approximate location relative to other features. Maxpooling (as well as other pooling methods) offer the benefit that there are many fewer pooled features, thus reducing the number of parameters needed in later layers of the CNN 900.

[0103] The final layer of connections in the network is a fully-connected layer that connects every7node from the pooling hidden layer 906 to every' one of the output nodes in the output layer 910. Using the example above, the input layer includes 28 x 28 nodes encoding the pixel intensities of the input image, the convolutional hidden layer 904 includes 3x24x24 hidden feature nodes based on application of a 5x5 local receptive field (for the filters) to three activation maps, and the pooling hidden layer 906 includes a layer of 3x 12x12 hidden feature nodes based on application of max-pooling filter to 2x2 regions across each of the three feature maps. Extending this example, the output layer 910 can include ten output nodes. In such an example, every node of the 3x12x12 pooling hidden layer 906 is connected to every7node of the output layer 910.

[0104] The fully connected layer 908 can obtain the output of the previous pooling hidden layer 906 (which should represent the activation maps of high-level features) and determines the features that most correlate to a particular class. For example, the fully- connected layer 908 can determine the high-level features that most strongly correlate to a particular class and can include weights (nodes) for the high-level features. A product can be computed between the w eights of the fully connected layer 908 and the pooling hidden layer 906 to obtain probabilities for the different classes. For example, if the CNN 900 is being used to predict that an object in an image is a person, high values will bePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO26 present in the activation maps that represent high-level features of people (e.g., two legs are present, a face is present at the top of the object, two eyes are present at the top left and top right of the face, a nose is present in the middle of the face, a mouth is present at the bottom of the face, and / or other features common for a person).

[0105] In some examples, the output from the output layer 910 can include an M- dimensional vector (in the prior example, M=10). M indicates the number of classes that the CNN 900 has to choose from when classifying the object in the image. Other example outputs can also be provided. Each number in the M-dimensional vector can represent the probability7the object is of a certain class. In one illustrative example, if a 10-dimensional output vector represents ten different classes of objects is [0 0 0.05 0.8 0 0.15 0 0 0 0], the vector indicates that there is a 5% probability that the image is the third class of object (e.g., a dog), an 80% probability that the image is the fourth class of object (e g., a human), and a 15% probability that the image is the sixth class of object (e.g., a kangaroo). The probability for a class can be considered a confidence level that the object is part of that class.

[0106] FIG. 10 is a block diagram of an example transformer in accordance with some aspects of the disclosure. For example, transformer 1000 may be an example of, or can implement, language models described herein.

[0107] In a convolutional neural network (CNN) model, the number of operations required to relate signals from two arbitrary7input or output positions grows in the distance between positions, which makes learning dependencies at different distant positions challenging for a CNN model. A transformer 1000 reduces the operations of learning dependencies by using an encoder 1010 and a decoder 1030 that implement an attention mechanism at different positions of a single sequence to compute a representation of that sequence. An attention function can be described as mapping a query and a set of keyvalue pairs to an output, where the query, keys, values, and output are all vectors. The output is computed as a weighted sum of the values, where the weight assigned to each value is computed by a compatibility7function of the query with the corresponding key.

[0108] In one example of a transformer, the encoder 1010 is composed of a stack of six identical layers and each layer has tw o sub-layers. The first sub-layer is a multi-head selfattention engine 1012, and the second sub-layer is a fully-connected feed-forwardPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO27 network 1014. A residual connection (not shown) connects around each of the sub-layers followed by normalization.

[0109] In this example transformer 1000, the decoder 1030 is also composed of a stack of six 6 identical layers. The decoder also includes a masked multi-head self-attention engine 1032, a multi-head attention engine 1034 over the output of the encoder 1010, and a fully-connected feed-forward network 1026. Each layer includes a residual connection (not shown) around the layer, which is followed by layer normalization. The masked multi -head self-attention engine 1032 is masked to prevent positions from attending to subsequent positions and ensures that the predictions at position i can depend only on the known outputs at positions less than i (e.g., auto-regression).

[0110] In the transformer, the queries, keys, and values are linearly projected by a multi -head attention engine into learned linear projects, and then attention is performed in parallel on each of the learned linear projects, which are concatenated and then projected into final values.

[0111] The transformer also includes a positional encoder 1040 to encode positions because the model does not contain recurrence and convolution and relative or absolute position of the tokens is needed. In the transformer 1000, the positional encodings are added to the input embeddings at the bottom layer of the encoder 1010 and the decoder 1030. The positional encodings are summed with the embeddings because the positional encodings and embeddings have the same dimensions. A corresponding position decoder 1050 is configured to decode the positions of the embeddings for the decoder 1030.

[0112] In some aspects, the transformer 1000 uses self-attention mechanisms to selectively weigh the importance of different parts of an input sequence during processing and allows the model to attend to different parts of the input sequence while generating the output. The input sequence is first embedded into vectors and then passed through multiple layers of self-attention and feed-forward networks. The transformer 1000 can process input sequences of variable length, making it well-suited for natural language processing tasks where input lengths can vary greatly. Additionally, the self-attention mechanism allows the transformer 1000 to capture long-range dependencies between words in the input sequence, which is difficult for RNNs and CNNs. The transformer with self-attention has achieved results in several natural language processing tasks that arePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO28 beyond the capabilities of other neural networks and has become a popular choice for language and text applications. For example, the various large language models, such as a generative pretrained transformer (e.g., ChatGPT, etc.) and other current models are types of transformer networks.

[0113] FIG. 11 illustrates an example computing-device architecture 1100 of an example computing device which can implement the various techniques described herein. In some examples, the computing device can include a mobile device, awearable device, an extended reality device (e.g., a virtual reality (VR) device, an augmented reality (AR) device, or a mixed reality (MR) device), a personal computer, a laptop computer, a video server, a vehicle (or computing device of a vehicle), or other device. For example, the computing-device architecture 1100 may include, implement, or be included in any or all of systems depicted in FIGs. 1-6, or by another system or device described herein. Additionally or alternatively, computing-device architecture 1100 may be configured to perform process 700, and / or other process described herein.

[0114] The components of computing-device architecture 1100 are show n in electrical communication with each other using connection 1112. such as a bus. The example computing-device architecture 1100 includes a processing unit (CPU or processor) 1102 and connection 1112 that couples various computing device components including computing device memory 1110, such as read only memory (ROM) 1108 and randomaccess memory (RAM) 1106, to processor 1102.

[0115] Computing-device architecture 1100 can include a cache of high-speed memory connected directly with, in close proximity to, or integrated as part of processor 1102. Computing-device architecture 1100 can copy data from memory 1110 and / or the storage device 11 14 to cache 1 104 for quick access by processor 1102. In this way, the cache can provide a performance boost that avoids processor 1102 delays while waiting for data. These and other modules can control or be configured to control processor 1102 to perform various actions. Other computing device memory 1110 may be available for use as well. Memory 11 10 can include multiple different types of memory with different performance characteristics. Processor 1102 can include any general-purpose processor and a hardw are or software service, such as service 1 1116, service 2 1118, and serv ice 3 1120 stored in storage device 1114. configured to control processor 1102 as well as aPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO29 special-purpose processor where software instructions are incorporated into the processor design. Processor 1102 may be a self-contained system, containing multiple cores or processors, a bus, memory' controller, cache, etc. A multi-core processor may be symmetric or asymmetric.

[0116] To enable user interaction with the computing-device architecture 1100, input device 1122 can represent any number of input mechanisms, such as a microphone for speech, a touch-sensitive screen for gesture or graphical input, keyboard, mouse, motion input, speech and so forth. Output device 1124 can also be one or more of a number of output mechanisms known to those of skill in the art, such as a display, projector, television, speaker device, etc. In some instances, multimodal computing devices can enable a user to provide multiple types of input to communicate with computing-device architecture 1100. Communication interface 1126 can generally govern and manage the user input and computing device output. There is no restriction on operating on any particular hardware arrangement and therefore the basic features here may easily be substituted for improved hardware or firmware arrangements as they are developed.

[0117] Storage device 1114 is a non-volatile memory and can be a hard disk or other types of computer readable media which can store data that are accessible by a computer, such as magnetic cassettes, flash memory cards, solid state memory' devices, digital versatile discs (DVDs), cartridges, random-access memories (RAMs) 1106, read only memory' (ROM) 1108, and hybrids thereof. Storage device 1114 can include services 1116. 1118, and 1120 for controlling processor 1102. Other hardware or software modules are contemplated. Storage device 1114 can be connected to the connection 1112. In one aspect, a hardware module that performs a particular function can include the software component stored in a computer-readable medium in connection with the necessary hardware components, such as processor 1102, connection 1112, output device 1124. and so forth, to carry out the function.

[0118] The term "‘substantially,’' in reference to a given parameter, property, or condition, may refer to a degree that one of ordinary skill in the art would understand that the given parameter, property, or condition is met with a small degree of variance, such as, for example, within acceptable manufacturing tolerances. By way of example, depending on the particular parameter, property, or condition that is substantially met, thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO30 parameter, property, or condition may be at least 90% met, at least 95% met, or even at least 99% met.

[0119] Aspects of the present disclosure are applicable to any suitable electronic device (such as security systems, smartphones, tablets, laptop computers, vehicles, drones, or other devices) including or coupled to one or more active depth sensing systems. While described below with respect to a device having or coupled to one light projector, aspects of the present disclosure are applicable to devices having any number of light projectors and are therefore not limited to specific devices.

[0120] The term '‘device’’ is not limited to one or a specific number of physical objects (such as one smartphone, one controller, one processing system and so on). As used herein, a device may be any electronic device with one or more parts that may implement at least some portions of this disclosure. While the below description and examples use the term “device” to describe various aspects of this disclosure, the term “device” is not limited to a specific configuration, type, or number of objects. Additionally, the term “system” is not limited to multiple components or specific aspects. For example, a system may be implemented on one or more printed circuit boards or other substrates and may have movable or static components. While the below description and examples use the term '‘system” to describe various aspects of this disclosure, the term “system” is not limited to a specific configuration, type, or number of objects.

[0121] Specific details are provided in the description above to provide a thorough understanding of the aspects and examples provided herein. However, it will be understood by one of ordinary skill in the art that the aspects may be practiced without these specific details. For clarity of explanation, in some instances the present technology may be presented as including individual functional blocks including functional blocks including devices, device components, steps or routines in a method embodied in software, or combinations of hardware and software. Additional components may be used other than those shown in the figures and / or described herein. For example, circuits, systems, networks, processes, and other components may be shown as components in block diagram form in order not to obscure the aspects in unnecessary detail. In other instances, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary' detail in order to avoid obscuring the aspects.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO31

[0122] Individual aspects may be described above as a process or method which is depicted as a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a block diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be re-arranged. A process is terminated when its operations are completed but could have additional steps not included in a figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination can correspond to a return of the function to the calling function or the main function.

[0123] Processes and methods according to the above-described examples can be implemented using computer-executable instructions that are stored or otherwise available from computer-readable media. Such instructions can include, for example, instructions and data which cause or otherwise configure a general-purpose computer, special purpose computer, or a processing device to perform a certain function or group of functions. Portions of computer resources used can be accessible over a network. The computer executable instructions may be, for example, binaries, intermediate format instructions such as assembly language, firmware, source code, etc.

[0124] The term '‘computer-readable medium’’ includes, but is not limited to, portable or non-portable storage devices, optical storage devices, and various other mediums capable of storing, containing, or carrying instruction(s) and / or data. A computer-readable medium may include a non-transitory medium in which data can be stored and that does not include carrier waves and / or transitory electronic signals propagating wirelessly or over wired connections. Examples of a non-transitory medium may include, but are not limited to, a magnetic disk or tape, optical storage media such as compact disk (CD) or digital versatile disk (DVD), flash memory, magnetic or optical disks, USB devices provided with non-volatile memory, networked storage devices, any suitable combination thereof, among others. A computer-readable medium may have stored thereon code and / or machine-executable instructions that may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a class, or any combination of instructions, data structures, or program statements. A code segment may be coupled to another code segment or a hardware circuit by passing and / or receivingPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO32 information, data, arguments, parameters, or memory contents. Information, arguments, parameters, data, etc. may be passed, forwarded, or transmitted via any suitable means including memory sharing, message passing, token passing, network transmission, or the like.

[0125] In some aspects the computer-readable storage devices, mediums, and memories can include a cable or wireless signal containing a bit stream and the like. However, when mentioned, non-transitory computer-readable storage media expressly exclude media such as energy, carrier signals, electromagnetic waves, and signals per se.

[0126] Devices implementing processes and methods according to these disclosures can include hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof, and can take any of a variety of form factors. When implemented in software, firmware, middleware, or microcode, the program code or code segments to perform the necessary tasks (e.g., a computer-program product) may be stored in a computer-readable or machine-readable medium. A processor(s) may perform the necessary tasks. Typical examples of form factors include laptops, smart phones, mobile phones, tablet devices or other small form factor personal computers, personal digital assistants, rackmount devices, standalone devices, and so on. Functionality described herein also can be embodied in peripherals or add-in cards. Such functionality can also be implemented on a circuit board among different chips or different processes executing in a single device, by way of further example.

[0127] The instructions, media for conveying such instructions, computing resources for executing them, and other structures for supporting such computing resources are example means for providing the functions described in the disclosure.

[0128] In the foregoing description, aspects of the application are described with reference to specific aspects thereof, but those skilled in the art will recognize that the application is not limited thereto. Thus, while illustrative aspects of the application have been described in detail herein, it is to be understood that the inventive concepts may be otherwise variously embodied and employ ed, and that the appended claims are intended to be construed to include such variations, except as limited by the prior art. Various features and aspects of the above-described application may be used individually or jointly. Further, aspects can be utilized in any number of environments and applicationsPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO33 beyond those described herein without departing from the broader spirit and scope of the specification. The specification and drawings are, accordingly, to be regarded as illustrative rather than restrictive. For the purposes of illustration, methods were described in a particular order. It should be appreciated that in alternate aspects, the methods may be performed in a different order than that described.

[0129] One of ordinary skill will appreciate that the less than (“<“) and greater than (“>“) symbols or terminology used herein can be replaced with less than or equal to (“<”) and greater than or equal to (“>”) symbols, respectively, without departing from the scope of this description.

[0130] Where components are described as being ‘'configured to” perform certain operations, such configuration can be accomplished, for example, by designing electronic circuits or other hardware to perform the operation, by programming programmable electronic circuits (e.g., microprocessors, or other suitable electronic circuits) to perform the operation, or any combination thereof.

[0131] The phrase “coupled to” refers to any component that is physically connected to another component either directly or indirectly, and / or any component that is in communication with another component (e.g., connected to the other component over a wired or wireless connection, and / or other suitable communication interface) either directly or indirectly.

[0132] Claim language or other language reciting “at least one of’ a set and / or “one or more” of a set indicates that one member of the set or multiple members of the set (in any combination) satisfy the claim. For example, claim language reciting “at least one of A and B” or “at least one of A or B” means A, B, or A and B. In another example, claim language reciting “at least one of A, B, and C” or “at least one of A, B, or C” means A, B, C, or A and B, or A and C, or B and C, A and B and C, or any duplicate information or data (e g., A and A, B and B, C and C, A and A and B, and so on), or any other ordering, duplication, or combination of A, B, and C. The language “at least one of’ a set and / or “one or more” of a set does not limit the set to the items listed in the set. For example, claim language reciting “at least one of A and B” or “at least one of A or B” may mean A, B, or A and B, and may additionally include items not listed in the set of A and B. The phrases “at least one” and “one or more” are used interchangeably herein.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO34

[0133] Claim language or other language reciting “at least one processor configured to,” “at least one processor being configured to,” “one or more processors configured to,” “one or more processors being configured to,” or the like indicates that one processor or multiple processors (in any combination) can perform the associated operation(s). For example, claim language reciting “at least one processor configured to: X. Y, and Z” means a single processor can be used to perform operations X, Y, and Z; or that multiple processors are each tasked with a certain subset of operations X, Y, and Z such that together the multiple processors perform X, Y, and Z; or that a group of multiple processors work together to perform operations X, Y, and Z. In another example, claim language reciting “at least one processor configured to: X, Y, and Z” can mean that any single processor may only perform at least a subset of operations X, Y, and Z.

[0134] Where reference is made to one or more elements performing functions (e.g., steps of a method), one element may perform all functions, or more than one element may collectively perform the functions. When more than one element collectively performs the functions, each function need not be performed by each of those elements (e.g., different functions may be performed by different elements) and / or each function need not be performed in whole by only one element (e.g., different elements may perform different sub-functions of a function). Similarly, where reference is made to one or more elements configured to cause another element (e.g.. an apparatus) to perform functions, one element may be configured to cause the other element to perform all functions, or more than one element may collectively be configured to cause the other element to perform the functions.

[0135] Where reference is made to an entity (e.g., any entity or device described herein) performing functions or being configured to perform functions (e.g., steps of a method), the entity may be configured to cause one or more elements (individually or collectively) to perform the functions. The one or more components of the entity may include at least one memory, at least one processor, at least one communication interface, another component configured to perform one or more (or all) of the functions, and / or any combination thereof. Where reference to the entity7performing functions, the entity7maybe configured to cause one component to perform all functions, or to cause more than one component to collectively perform the functions. When the entity7is configured to causePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO35 more than one component to collectively perform the functions, each function need not be performed by each of those components (e.g., different functions may be performed by different components) and / or each function need not be performed in whole by only one component (e.g., different components may perform different sub-functions of a function).

[0136] The various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the aspects disclosed herein may be implemented as electronic hardware, computer software, firmware, or combinations thereof. To clearly illustrate this interchangeability7of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality- . Whether such functionality7is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality7in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

[0137] The techniques described herein may also be implemented in electronic hardware, computer software, firmware, or any combination thereof. Such techniques may be implemented in any of a variety of devices such as general-purposes computers, wireless communication device handsets, or integrated circuit devices having multiple uses including application in wireless communication device handsets and other devices. Any features described as modules or components may be implemented together in an integrated logic device or separately as discrete but interoperable logic devices. If implemented in software, the techniques may be realized at least in part by a computer- readable data storage medium including program code including instructions that, when executed, performs one or more of the methods described above. The computer-readable data storage medium may form part of a computer program product, which may include packaging materials. The computer-readable medium may include memory or data storage media, such as random-access memory7(RAM) such as synchronous dynamic random-access memory (SDRAM), read-only memory7(ROM), non-volatile randomaccess memory (NVRAM), electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic or optical data storage media, and the like. ThePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO36 techniques additionally, or alternatively, may be realized at least in part by a computer- readable communication medium that carries or communicates program code in the form of instructions or data structures and that can be accessed, read, and / or executed by a computer, such as propagated signals or waves.

[0138] The program code may be executed by a processor, which may include one or more processors, such as one or more digital signal processors (DSPs), general-purpose microprocessors, an application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuitry. Such a processor may be configured to perform any of the techniques described in this disclosure. A general-purpose processor may be a microprocessor; but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, such as, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. Accordingly, the term ‘“processor as used herein may refer to any of the foregoing structure, any combination of the foregoing structure, or any other structure or apparatus suitable for implementation of the techniques described herein.

[0139] Illustrative aspects of the disclosure include:

[0140] Aspect 1. An apparatus for detecting inconsistencies between outputs of a machine learning model, the apparatus comprising: one or more memories configured to store input data; and one or more processors coupled to the one or more memories and configured to: process, using the machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.

[0141] Aspect 2. The apparatus of Aspect 1, wherein the one or more processors are configured to: determine a similarity score representing a similarity between the firstPolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO37 output and the second output; and determine that the contradiction exists further based on the similarity score.

[0142] Aspect 3. The apparatus of Aspect 2, wherein, to determine the similarity score, the one or more processors are configured to: determine a first text embedding of the first output; determine a second text embedding of the second output; and determine the similarity score based on a comparison between the first text embedding and the second text embedding.

[0143] Aspect 4. The apparatus of any of Aspects 2 or 3, wherein, to determine the similarity score, the one or more processors are configured to: process the first output and the second output using an additional machine learning model to determine the similarity score.

[0144] Aspect 5. The apparatus of Aspects 4, wherein the additional machine learning model is trained to detect textual similarities.

[0145] Aspect 6. The apparatus of any of Aspects 1-5, wherein the one or more processors are configured to: determine a similarity score representing a similarity between the first output and the second output; compare the similarity score to a similarity threshold and the contradiction score to a contradiction threshold; and determine that the contradiction exists based on a determination that the similarity score is greater than the similarity threshold and a determination that the contradiction score is greater than the contradiction threshold.

[0146] Aspect 7. The apparatus of any of Aspects 1-6, wherein, to determine the contradiction score, the one or more processors are configured to: compare the first output and the second output to determine at least one of an entailment value, a contradiction value, or a neutrality value; and determine the contradiction score as the contradiction value.

[0147] Aspect 8. The apparatus of any of Aspects 1-7, wherein, to adjust the first output, the one or more processors are configured to: process, using the machine learning model, the first output and a prompt to generate the adjusted first output, wherein the prompt includes instructions for the machine learning model to revise the first output.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO38

[0148] Aspect 9. The apparatus of any of Aspects 1-8, wherein the one or more processors are configured to determine that the contradiction exists is further based on a user input indicative of the contradiction received from a user device.

[0149] Aspect 10. The apparatus of any of Aspects 1-9, wherein, to determine the contradiction exists, the one or more processors are configured to: process, using an additional machine learning model, the first output and the second output to generate an indication of a contradiction between the first output and the second output, wherein the additional machine learning model is trained to detect textual similarities.

[0150] Aspect 11. The apparatus of any of Aspects 1-10, wherein the one or more processors are configured to: determine one or more additional contradictions between at least two outputs; determine that a number of the one or more additional contradictions is greater than an additional threshold; and reset the machine learning model based on the number of the one or more additional contradictions being greater than the additional threshold.

[0151] Aspect 12. The apparatus of any of Aspects 1-1 1, wherein the one or more processors are configured to: provide the adjusted first output and the second output to a user device.

[0152] Aspect 13. The apparatus of any of Aspects 1-12, wherein the first input comprises a first text embedding associated with a first question, the first output comprises text associated with a first answer to the first question, the second input comprises a second text embedding associated with a second question, and the second output comprises text associated with a second answer to the second question.

[0153] Aspect 14. The apparatus of any of Aspects 1-13, further comprising one or more displays configured to display the adjusted first output and the second output.

[0154] Aspect 15. A method comprising: processing, using a machine learning model, a first input to generate a first output; processing, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determining that a contradiction exists between the first output and the second output based on thePolsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO39 contradiction score; and adjusting the first output based on a determination the contradiction exists between the first output and the second output.

[0155] Aspect 16. The method of Aspect 15, further comprising: determining a similarity score representing a similarity between the first output and the second output; and determining that the contradiction exists further based on the similarity score.

[0156] Aspect 17. The method of Aspect 16, further comprising: determining a first text embedding of the first output; determining a second text embedding of the second output; and determining the similarity score based on a comparison between the first text embedding and the second text embedding.

[0157] Aspect 18. The method of any of Aspects 16 or 17, wherein determining the similarity score comprises: processing the first output and the second output using an additional machine learning model to determine the similarity' score.

[0158] Aspect 19. The method of any of Aspects 15-18, further comprising: determining a similarity' score representing a similarity' between the first output and the second output; comparing the similarity score to a similarity threshold and the contradiction score to a contradiction threshold; and determining that the contradiction exists based on a determination that the similarity score is greater than the similarity threshold and a determination that the contradiction score is greater than the contradiction threshold.

[0159] Aspect 20. The method of any of Aspects 15-19, wherein determining the contradiction score comprises: comparing the first output and the second output to determine at least one of an entailment value, a contradiction value, or a neutrality value; and determining the contradiction score as the contradiction value.

[0160] Aspect 21. The method of any of Aspects 15-20, wherein adjusting the first output comprises: process, using the machine learning model, the first output and a prompt to generate the adjusted first output, wherein the prompt includes instructions for the machine learning model to revise the first output.

[0161] Aspect 22. The method of any of Aspects 15-11, further comprising determining that the contradiction exists is further based on a user input indicative of the contradiction received from a user device.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO40

[0162] Aspect 23. The method of any of Aspects 15-22, wherein determining that the contradiction exists comprises: process, using an additional machine learning model, the first output and the second output to generate an indication of a contradiction between the first output and the second output, wherein the additional machine learning model is trained to detect textual similarities.

[0163] Aspect 24. The method of any of Aspects 15-23. further comprising: determine one or more additional contradictions between at least two outputs; determine that a number of the one or more additional contradictions is greater than an additional threshold; and reset the machine learning model based on the number of the one or more additional contradictions being greater than the additional threshold.

[0164] Aspect 25. The method of any of Aspects 15-24, further comprising: provide the adjusted first output and the second output to a user device.

[0165] Aspect 26. The method of any of Aspects 15-25, wherein the first input comprises a first text embedding associated with a first question, the first output comprises text associated with a first answer to the first question, the second input comprises a second text embedding associated with a second question, and the second output comprises text associated with a second answer to the second question.

[0166] Aspect 27. A non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to perform operations according to any of Aspects 15-26.

[0167] Aspect 28. An apparatus including one or more means for performing operations according to any of Aspects 15-26.Polsinelli Ref. No 094922-863924

Claims

Qualcomm Ref. No. 2500985WO41CLAIMSWHAT IS CLAIMED IS:

1. An apparatus for detecting inconsistencies between outputs of a machine learning model, the apparatus comprising: one or more memories configured to store input data; and one or more processors coupled to the one or more memories and configured to: process, using the machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.

2. The apparatus of claim 1, wherein the one or more processors are configured to: determine a similarity score representing a similarity’ between the first output and the second output; and determine that the contradiction exists further based on the similarity score.

3. The apparatus of claim 2, wherein, to determine the similarity score, the one or more processors are configured to: determine a first text embedding of the first output; determine a second text embedding of the second output; and determine the similarity' score based on a comparison between the first text embedding and the second text embedding.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO424. The apparatus of claim 2, wherein, to determine the similarity score, the one or more processors are configured to: process the first output and the second output using an additional machine learning model to determine the similarity score.

5. The apparatus of claim 4, wherein the additional machine learning model is trained to detect textual similarities.

6. The apparatus of claim 1, wherein the one or more processors are configured to: determine a similarity score representing a similarity between the first output and the second output; compare the similarity score to a similarity threshold and the contradiction score to a contradiction threshold; and determine that the contradiction exists based on a determination that the similarity score is greater than the similarity threshold and a determination that the contradiction score is greater than the contradiction threshold.

7. The apparatus of claim 1, wherein, to determine the contradiction score, the one or more processors are configured to: compare the first output and the second output to determine at least one of an entailment value, a contradiction value, or a neutrality value; and determine the contradiction score as the contradiction value.

8. The apparatus of claim 1 , wherein, to adjust the first output, the one or more processors are configured to: process, using the machine learning model, the first output and a prompt to generate the adjusted first output, wherein the prompt includes instructions for the machine learning model to revise the first output.

9. The apparatus of claim 1, wherein the one or more processors are configured to determine that the contradiction exists is further based on a user input indicative of the contradiction received from a user device.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO4310. The apparatus of claim 1, wherein, to determine the contradiction exists, the one or more processors are configured to: process, using an additional machine learning model, the first output and the second output to generate an indication of a contradiction between the first output and the second output, wherein the additional machine learning model is trained to detect textual similarities.

11. The apparatus of claim 1, wherein the one or more processors are configured to: determine one or more additional contradictions between at least two outputs: determine that a number of the one or more additional contradictions is greater than an additional threshold; and reset the machine learning model based on the number of the one or more additional contradictions being greater than the additional threshold.

12. The apparatus of claim 1, wherein the one or more processors are configured to: provide the adjusted first output and the second output to a user device.

13. The apparatus of claim 1, wherein the first input comprises a first text embedding associated with a first question, the first output comprises text associated with a first answer to the first question, the second input comprises a second text embedding associated with a second question, and the second output comprises text associated with a second answer to the second question.

14. The apparatus of claim 1, further comprising one or more displays configured to display the adjusted first output and the second output.

15. A method comprising: processing, using a machine learning model, a first input to generate a first output; processing, using the machine learning model, a second input to generate a second output;Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO44 determine a contradiction score representing a contradiction between the first output and the second output; determining that a contradiction exists between the first output and the second output based on the contradiction score; and adjusting the first output based on a determination the contradiction exists between the first output and the second output.

16. The method of claim 15, further comprising: determining a similarity score representing a similarity between the first output and the second output; and determining that the contradiction exists further based on the similarity score.

17. The method of claim 16, further comprising: determining a first text embedding of the first output; determining a second text embedding of the second output; and determining the similarity score based on a comparison between the first text embedding and the second text embedding.

18. The method of claim 16, wherein determining the similarity score comprises: processing the first output and the second output using an additional machine learning model to determine the similarity' score.

19. The method of claim 15, further comprising: determining a similarity score representing a similarity between the first output and the second output; comparing the similarity score to a similarity' threshold and the contradiction score to a contradiction threshold; and determining that the contradiction exists based on a determination that the similarity' score is greater than the similarity7threshold and a determination that the contradiction score is greater than the contradiction threshold.Polsinelli Ref. No 094922-863924Qualcomm Ref. No. 2500985WO4520. A non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to: process, using a machine learning model, a first input to generate a first output; process, using the machine learning model, a second input to generate a second output; determine a contradiction score representing a contradiction between the first output and the second output; determine that a contradiction exists between the first output and the second output based on the contradiction score; and adjust the first output based on a determination the contradiction exists between the first output and the second output.Polsinelli Ref. No 094922-863924

Images