Biometric activity application for measuring a fatigue state

Abstract

A computing system for determining a fatigue level or fatigue state includes one or more memories to store instructions and one or more processors to execute the instructions to perform operations, the operations including: receiving a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

Description

PCT / US24 / 54882 07 November 2024 (07.11.2024)BIOMETRIC ACTIVITY APPLICATION FOR MEASURING A FATIGUE STATEFIELD

[0001] This disclosure relates generally to computing devices. More particularly, the disclosure relates to computing devices which are used to obtain and analyze biometric information of a user to determine information about a user including a fatigue state or fatigue level of the user.BACKGROUND

[0002] Existing wearable computing devices include sensors that can provide information for tracking the sleep quality of a user. Metrics that can be measured by the wearable computing devices can include sleep duration, sleep stages, and night awakenings. While these sensors provide useful data, they are limited in how they inform the user. Data related to sleep quality may be presented as time-series charts, placing the burden on the user to interpret and track their own data. It is also often not clear how such data correlates or affects the daytime performance of the user, any health symptoms the user is experiencing, or other activities of the user such as exercise.SUMMARY

[0003] Aspects and advantages of embodiments of the disclosure will be set forth in part in the following description, or can be learned from the description, or can be learned through practice of the embodiments.

[0004] Example aspects of the disclosure provide an example computing system or computing device (e.g., a user computing device) that includes one or more processors and one or more example non-transitory computer-readable media storing instructions that are executable by the one or more processors to cause the computing system or device to perform example operations. In some implementations, the example operations can include receiving a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0005] In some implementations, the operations further comprise obtaining the context information from the voice input, and obtaining the context information from the voice input comprises: obtaining a transcription of the voice input; and processing, by one or more second machine-learned models, the transcription, to output a textual summary of the voice input, and wherein the context information includes the textual summary.

[0006] In some implementations, obtaining the context information from the voice input comprises: processing, by one or more third machine-learned models, the transcription, to output at least one of one or more topics or one or more sentiments associated with the voice input, and wherein the context information includes the at least one of the one or more topics or the one or more sentiments.

[0007] In some implementations, the operations further comprise obtaining the audio biomarkers from the audio information, and obtaining the audio biomarkers from the audio information comprises: obtaining a transcription of the voice input and associating a timestamp with each word utterance in the transcription; and determining, based on the transcription and the timestamp associated with each word utterance in the transcription, a plurality of time-based features, and wherein the audio biomarkers include the plurality of time-based features.

[0008] In some implementations, the plurality of time-based features include at least one of a total length of speech, total number of words spoken, rate of speech, speech percentage, mean pause time between words, or pause variability.

[0009] In some implementations, the operations further comprise obtaining the audio biomarkers from the audio information, and obtaining the audio biomarkers from the audio information comprises: processing, by one or more second machine-learned models, raw audio associated with the voice input, to determine a fatigue indication, and wherein the audio biomarkers include the fatigue indication.

[0010] In some implementations, processing, by the one or more second machine-learned models, the raw audio associated with the voice input, to determine the fatigue indication comprises extracting a plurality of acoustic-based features, and the operations further comprise applying a similarity function to the plurality of acoustic-based features and a plurality of baseline acoustic-based features, to determine the fatigue indication.

[0011] In some implementations, the plurality of acoustic-based features related to at least one of vocal patterns, tone, or intonations associated with the voice input of the user.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0012] In some implementations, the operations further comprise processing, by one or more second machine-learned models, the fatigue level, to generate content including a recommendation to reduce the fatigue level associated with the user.

[0013] In some implementations, the operations further comprise: retrieving historical information associated with the user, the historical information including a plurality of prior fatigue levels from prior time periods and sleep quality information associated with the prior time periods; and processing, by one or more second machine-learned models, the fatigue level and the historical information, to generate content including a recommendation to reduce a sleep debt associated with the user.

[0014] In some implementations, the operations further comprise receiving sensor data including biometric information of the user, and processing, by the one or more first machine-learned models, the sensor data, the first fatigue information, and the second fatigue information, to output the fatigue level associated with the user.

[0015] In some implementations, the biometric information of the user includes at least one of total sleep time, sleep onset latency, sleep efficiency, sleep stage information, number of awakenings, or sleep disturbances.

[0016] Example aspects of the disclosure provide an example computer-implemented method. In some implementations, the example computer-implemented method can include: receiving, by a computing system comprising one or more processors and one or more first machine-learned models, a voice input from a user describing a biometnc state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by the one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

[0017] In some implementations, the method includes obtaining the context information from the voice input by: obtaining a transcription of the voice input; and processing, by one or more second machine-learned models, the transcription, to output a textual summary describing the biometric state of the user, and wherein the context information includes the textual summary.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0018] In some implementations, the method includes processing, by one or more third machine-learned models, the transcription, to output at least one of one or more topics or one or more sentiments associated with the voice input, and wherein the context information includes the at least one of the one or more topics or the one or more sentiments.

[0019] In some implementations, the method includes obtaining the audio biomarkers from the audio information by: obtaining a transcription of the voice input and associating a timestamp with each word utterance in the transcription; and determining, based on the transcription and the timestamp associated with each word utterance in the transcription, a plurality of time-based features, and wherein the audio biomarkers include the plurality of time-based features.

[0020] In some implementations, the method includes obtaining the audio biomarkers from the audio information by: processing, by one or more second machine-learned models, raw audio associated with the voice input, to determine a fatigue indication, and wherein the audio biomarkers include the fatigue indication.

[0021] In some implementations, the method includes receiving sensor data including biometric information of the user, the context information includes a textual summary of the voice input, topics associated with the voice input, and sentiments associated with the voice input, the audio biomarkers include time-based features associated with the voice input and acoustic-based features related to at least one of vocal patterns, tone, or intonations associated with the voice input of the user, and processing, by the one or more first machine-learned models, the sensor data, the first fatigue information, and the second fatigue information, to output the fatigue level associated with the user.

[0022] In some implementations, the method includes retrieving historical information associated with the user, the historical information including a plurality of prior fatigue levels from prior time periods and sleep quality information associated with the prior time periods; and processing, by one or more second machine-learned models, the fatigue level and the historical information, to generate content including a recommendation to reduce a sleep debt associated with the user.

[0023] The computer-implemented method may execute any of the operations of the computing systems or computing devices as described herein.

[0024] Example aspects of the disclosure provide one or more example non-transitory computer-readable media storing instructions that are executable by one or more processorsPCT / US24 / 54882 07 November 2024 (07.11.2024) to cause a computing system to perform example operations. In some implementations, the example operations can include receiving a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

[0025] The non-transitory computer-readable medium may store additional instructions to execute other aspects and operations of the computing systems or computing devices and computer-implemented methods as described herein.

[0026] Other example aspects of the disclosure are directed to other systems, methods, apparatuses, tangible non-transitory computer-readable media, and devices for performing functions described herein. These and other features, aspects, and advantages of various implementations will become better understood with reference to the following description and appended claims. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate implementations of the disclosure and, together with the description, help explain the related principles.BRIEF DESCRIPTION OF THE DRAWINGS

[0027] FIG. 1 A is an example system, according to one or more example embodiments of the disclosure;

[0028] FIG. IB is an example block diagram of a computing system, according to one or more example embodiments of the disclosure;

[0029] FIGS. 2A-2C each illustrate a flow diagram of an example, non-limiting computer- implemented method, according to one or more example embodiments of the disclosure;

[0030] FIG. 3 illustrates an example block diagram of a system including a biometric activity application, according to one or more example embodiments of the disclosure;

[0031] FIG. 4 illustrates an example block diagram of a system including one or more machine-learned models, according to one or more example embodiments of the disclosure;

[0032] FIGS. 5A-5B are example spectrograms of voice inputs, according to one or more example embodiments of the disclosure;PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0033] FIG. 6 is a flow chart diagram illustrating an example method for training a machine- learned model according to example implementations of aspects of the disclosure;

[0034] FIG. 7 is a block diagram of an example processing flow for using machine-learned model(s) to process input(s) to generate output(s) according to example implementations of aspects of the disclosure;

[0035] FIG. 8 is a block diagram of an example sequence processing model according to example implementations of aspects of the disclosure;

[0036] FIG. 9 is a block diagram of an example technique for populating an example input sequence for processing by a sequence processing model according to example implementations of aspects of the disclosure;

[0037] FIG. 10 is a block diagram of an example model development platform according to example implementations of aspects of the disclosure;

[0038] FIG. 11 is a block diagram of an example training workflow for training a machine- learned model according to example implementations of aspects of the disclosure;

[0039] FIG. 12 is a block diagram of an inference system for operating one or more machine- learned model(s) to perform inference according to example implementations of aspects of the disclosure;

[0040] FIG. 13 is a block diagram of an example networked computing system according to example implementations of aspects of the disclosure;

[0041] FIG. 14 is a block diagram of an example computing device according to example implementations of aspects of the disclosure; and

[0042] FIG. 15 is a block diagram of an example computing device according to example implementations of aspects of the disclosure.DETAILED DESCRIPTION

[0043] Reference now will be made to embodiments of the disclosure, one or more examples of which are illustrated in the drawings, wherein like reference characters denote like elements. Each example is provided by way of explanation of the disclosure and is not intended to limit the disclosure. In fact, it will be apparent to those skilled in the art that various modifications and variations can be made to disclosure without departing from the scope or spirit of the disclosure. For instance, features illustrated or described as part of onePCT / US24 / 54882 07 November 2024 (07.11.2024) embodiment can be used with another embodiment to yield a still further embodiment. Thus, it is intended that the disclosure covers such modifications and variations as come within the scope of the appended claims and their equivalents.

[0044] Sleep is an important activity that is vital for humans and is often overlooked due to the stresses and needs of daily life. Some studies have found that the prevalence of poor sleep in the general population is about 38%, which includes general lack of sleep and sleep disorders such as insomnia. Sleep deprivation is linked to a variety of health impacts including cognitive disorders, obesity , immune response, and stress I mental wellbeing. While these impacts can cause severe impacts to quality of life, most can be avoided through simple lifestyle changes and forming habits to encourage good “sleep hygiene.”

[0045] According to examples of the disclosure, computing systems and methods described herein can enable a user to understand their state of fatigue and alertness based on information including what the user says, how the user sounds when speaking, and the quality of their sleep (e.g., obtained from a phone, a wearable computing device such as a sleep tracker, etc.). The computing systems and methods described herein leverage one or more machine-learned models (e.g., one or more large language models) to understand both the content of a user’s sleep logs and, in some implementations in combination with other trained speech models, whether the user sounds fatigued or tired when completing the sleep log. This information can be collected and provided as an input to the one or more machine- learned models which may be part of an activity agent (e.g., a sleep activity agent which may act as a “sleep coach”) which can provide, as an output, an accurate assessment of a fatigue state of a user and can offer personalized advice or instructions to the user to help the user minimize daytime sleepiness, optimize their sleep and / or activity, and help achieve an improved quality of life.

[0046] According to examples of the disclosure, computing systems described herein are configured to receive various inputs including a voice input from a user describing how the user feels, for example, their level of alertness or how the user feels about their sleep or energy state (e.g., at a particular time of day). For example, the user can provide a voice input such as “I am having trouble focusing during meetings, and I woke up really tired today.” The computing system may also be configured to receive other inputs including sensor data which may include biometric information associated with the user (e.g., the duration of sleep the user, sleep stage information, information about night awakenings, blood pressure information, stress level information, etc.).PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0047] According to examples of the disclosure, the computing systems described herein are configured to, while the user is speaking, generate or extract a transcript of the speech of the user from the voice input. For example, a computing system may include an on-device speech transcriber which may be configured to extract the transcript of the user’s speech and label each word utterance with a precise timestamp (e.g., a millisecond timestamp).

[0048] According to examples of the disclosure, the computing systems described herein are configured to generate text embeddings based on the raw audio associated with the voice input and / or the transcribed text, and one or more topics associated with the voice input can be identified (labelled), and a summary of the voice input may be generated. For example, the computing system may be configured to implement one or more machine-learned models (e.g., one or more large language models) which are configured to label the one or more topics based on the generated text embeddings. For example, the one or more machine- learned models (e.g., one or more large language models) may be configured to interpret the content of the user’s voice input to generate the summary of the voice input. In some implementations, the summary may be associated with a sleep journal or sleep log of the user. In some implementations, the computing system may be configured to provide the transcript as an input to one or more instruction-tuned large language models to generate the summary for the sleep journal or sleep log, and to provide the transcript as an input to one or more fine-tuned stand-alone text classification models which is configured to extract topic and sentiment information relating to the voice input.

[0049] According to examples of the disclosure, the computing systems described herein are configured to extract biometric information (e.g., audio biomarkers) associated with the raw audio of the voice input and / or the transcript of the voice input. The computing system may be configured to collect (or determine) the audio biomarkers from the voice input and use the audio biomarkers to infer a state of the user (e.g., the user sounds tired, energetic, normal, abnormal, etc.). In some implementations, the computing system may confirm (verify) or seek feedback from the user regarding the determination as to the state of the user, and the computing system may be configured to request information from the user regarding a possible cause for the state of the user (e.g., whether the user did not get enough sleep, whether the user consumed certain foods or beverages, whether the user exercised or engaged in a particular biometric activity, etc.).

[0050] In some implementations, various features (audio biomarkers) can be extracted from the voice input by the computing system. For example, the features can include timing-based features, audio embeddings, and labels. For example, the timing-based features can includePCT / US24 / 54882 07 November 2024 (07.11.2024) the total length of speech, total number of words spoken, rate of speech in words per minute, speech percentage (e.g., time spoken over total time), mean pause time between words (e.g., in seconds), and pause variability (e.g., in seconds). The timing-based features can provide insight regarding a fatigue state of the user and can be utilized by the computing system as an input (e.g., to one or more machine-learned models) for determining a level of fatigue associated with the user.

[0051] For example, the features (audio biomarkers) can include acoustic-based features (e.g., audio embeddings) that can be extracted from the voice input (e.g., from the raw audio) by the computing system, which can be utilized to detect changes in speech or vocal patterns of the user, tone, intonations, and speaker identification information. In some implementations, the computing system may be configured to implement a similarity function between embeddings associated with the voice input and embeddings associated with other previously obtained audio information (e.g., a baseline audio input) to determine differences in the speech which can indicate a level of fatigue and / or differences in the voice of the user.

[0052] According to examples of the disclosure, computing systems and methods described herein can provide an accurate assessment of the fatigue state or alertness level of the user, for example, based on context information associated with a voice input, how a user sounds when they are speaking, and the qualify of the sleep of a user (e.g., based on sensor data obtained from a smartphone or wearable sleep tracker). The computing system can include a biometric application having an activity agent (e.g., sleep agent) that can act as a “sleep coach” to provide personalized and specific advice to the user to help the user minimize daytime sleepiness, optimize their sleep, optimize other biometric activities that may affect the user’s energy state, and help the user achieve an improved quality of life.

[0053] One or more technical benefits of the disclosure include the implementation of machine-learned models which generate dialogue for communications with a user and provide recommendations relating to a biometric activity (reducing fatigue, increasing energy levels, improving sleep qualify, reducing a sleep debt, etc.) based on biometric information associated with the user and dialogue information received from the user via a voice input. The computing systems and methods described herein improve the qualify and personalization of dialogue between the user and the computing systems as well as improve the qualify' of biometric assessments (e.g., fatigue assessments) by extracting context information from the voice input and inferring information about the user (e.g., a user’s health or physical or mental well-being, sentiment, emotional state, etc.). The computing systems and methods described herein improve the qualify and personalization of dialoguePCT / US24 / 54882 07 November 2024 (07.11.2024) between the user and the computing systems as well as improve the quality of biometric assessments (e.g., fatigue assessments) by extracting context information from the voice input and inferring information (e.g., atopic of the voice input, a summary of the voice input, etc.). The computing systems and methods described herein improve the quality and personalization of dialogue between the user and the computing systems as well as improve the quality' of biometric assessments (e.g., fatigue assessments) by extracting audio biomarkers from the voice input and inferring information (e.g., a fatigue level inference). Therefore, a technical effect achieved by the computing systems and methods described herein includes improving determinations regarding a fatigue state of the user and improving a user experience and quality of life based on recommendations and insights that are generated via one or more machine-learned models. The computing systems and methods described herein also improve the quality and personalization of insights and recommendations to the user by extracting context information and audio biomarkers from the voice input, and by further taking into account biometric information associated with the user.

[0054] The machine-learned models described herein can conserve computing resources including processing power, memory, network resources (e.g., bandwidth), etc., by interacting with the user in a more personalized manner and by providing accurate and personalized recommendations, reducing the need for additional requests by the user for revised recommendations, and saving time and computing resources by not requiring the user to input additional prompts or edit existing prompts and thus avoiding the need for processing prompts and generating further inferences. The machine-learned models described herein can conserve computing resources including network resources (e.g., bandwidth) and can improve the security of data associated with the user, for example, by implementing one or more machine-learned models on-device (e.g., at the user computing device) rather than at a remote device (e.g., at a server computing system). Further, in some implementations lighter machine-learned models (e.g., requiring less processing power and / or using a lower number of parameters) can be used for some operations such as classifying information in a voice input to conserve computing resources, compared to other machine-learned models which are used for other tasks (e.g., generating a textual summary of the voice input or generating recommendations or insights). Further, in some implementations the machine-learned models described herein can be embodied by pre-existing machine-learned models that are capable of processing prompts as described herein to generate the dialogue, insights, andPCT / US24 / 54882 07 November 2024 (07.11.2024) recommendations for interacting with the user as well as for generating outputs including a fatigue output indicating a fatigue state or fatigue level of the user. For example, enabling the reuse of a pre-existing machine-learned model with the new techniques described herein, can save or conserve storage on a computing device and / or time for training because it is not necessary to train and store a new model.

[0055] Therefore, aspects of the disclosure provide technical effects, benefits, and / or improvements in computing technology and the technology of biometric application systems, via one or more computing devices (e.g., a user computing device, a server computing system, and combinations thereof), as described herein.

[0056] Referring now to the drawings, FIG. 1 A is an example system according to one or more example embodiments of the disclosure. FIG. 1 A illustrates an example of a system 1100 which includes a computing device 100, an external computing device 200, a server computing system 300, and external content 500, which may be in communication with one another over a network 400. For example, the computing device 100 and the external computing device 200 can include any of a personal computer, a smartphone, a tablet computer, a laptop, a global positioning service device, a smartwatch, fitness tracker, and the like. The network 400 may include any type of communications network including a wired or wireless network, or a combination thereof The network 400 may include a local area network (LAN), wireless local area network (WLAN), wide area network (WAN), personal area network (PAN), virtual private network (VPN), or the like. For example, wireless communication between elements of the example embodiments may be performed via a wireless LAN, Wi-Fi, Bluetooth, ZigBee, Wi-Fi direct (WFD), ultra wideband (UWB), infrared data association (IrDA), Bluetooth low energy (BLE), near field communication (NFC), a radio frequency (RF) signal, and the like. For example, wired communication between elements of the example embodiments may be performed via a pair cable, a coaxial cable, an optical fiber cable, an Ethernet cable, and the like. Communication over the network 400 can use a wide variety of communication protocols (e.g., TCP / IP, HTTP, SMTP, FTP), encodings or formats (e.g., HTML, XML), and / or protection schemes (e.g., VPN, secure HTTP, SSL).

[0057] As will be explained in more detail below, in some implementations the computing device 100 and / or server computing system 300 may form part of an application system which can determine a fatigue state or fatigue level of a user and which can provide a tool for an activity recommendation system (e.g., biometric activity recommendation system) byPCT / US24 / 54882 07 November 2024 (07.11.2024) which, via machine-learned models described herein provide content and recommendations concerning a biometric activity to a user (e.g., concerning the sleep of a user).

[0058] In some example embodiments, the server computing system 300 may obtain data from one or more of a sensor data store 340, a biometric activity data store 350, a content data store 360, and a machine-learned model data store 370, to implement various operations and aspects of the application systems as disclosed herein. The sensor data store 340, biometric activity data store 350, content data store 360, and machine-learned model data store 370 may be integrally provided with the server computing system 300 (e.g., as part of the one or more memory devices 320 of the server computing system 300) or may be separately (e.g., remotely) provided. Further, sensor data store 340, biometric activity data store 350, content data store 360, and machine-learned model data store 370 can be combined as a single data store (database) or may include a plurality of respective data stores. Data stored in one data store (e.g., the biometric activity data store 350) may overlap with some data stored in another data store (e.g., sensor data store 340). In some implementations, one data store (e.g., the machine-learned model data store 370) may reference data that is stored in another data store (e.g., the sensor data store 340).

[0059] In some implementations, the sensor data store 340 can store information relating to information collected via one or more sensors. For example, the sensor data store 340 can store sensor data related to various biometrics, including biometrics associated with an electrocardiogram (ECG), photoplethysmography (PPG) information, heart rate, heart rate recovery', pulse information, body mass index information, heart rate variability, blood pressure, oxygen saturation, body temperature, sleep metrics (e.g., sleep scores, total sleep time, sleep onset latency, sleep efficiency, sleep stage information, REM sleep latency, number of awakenings, sleep disturbances, etc.), physical activities (e.g., number of steps walked, number of miles cycled, number of laps swam, etc.), audio spectrograms, and the like. For example, the sensor data store 340 can store sensor data related to other information including imagery and / or videos captured by image sensors, lighting information, weather information, noise information, movement or motion information, etc.

[0060] In some implementations, the information stored in the sensor data store 340 can be associated with and / or stored according to a particular user or a plurality of users, according to a particular sensor category, particular biometric activity (e.g., sleep activity), sensor context, according to a particular time, location, content type, etc. In some implementations, the information stored in the sensor data store 340 can be associated with and / or storedPCT / US24 / 54882 07 November 2024 (07.11.2024) according to a particular environment (e.g., outdoor, indoor, etc.). In some implementations, the information stored in the sensor data store 340 can be associated with and / or stored according to a particular entity that is associated with the collection of the sensor data (e.g., an entity that collected the sensor data, a user that is associated with the sensor data, an entity to which the collected sensor data is transmitted, etc.). For example, sensor data may be stored or retrieved from the sensor data store 340 that is relevant to a dialogue between a user and an activity agent for determining a fatigue state or fatigue level of a user, providing content, generating dialogue for communication with a user, or providing a recommendation to a user.

[0061] In some implementations, the biometric activity data store 350 can store information relating to biometric activities that are performed by a user. The biometric activity data store 350 can store information relating to particular biometric activities (e.g., sleep activities, exercise activities, nutrition activities, etc.) and may include goals associated with the biometric activity and the user. The information relating to the particular biometric activities can include a time engaged in performing the biometric activity, a metric or score relating to the performance of the biometric activity (e.g., a distance travelled, a number of laps swam, miles bicycled, repetitions and / or sets performed, calories eaten, hours slept, etc.). The information relating to the particular biometric activities can also include or be associated with biometric information collected during the biometric activity (e.g., a heart rate measured while engaged in performing the biometric activity, a stress level, a detected sleep state such as REM, Nl, N2, and N3 stages, calories burned, etc.).

[0062] In some implementations, the information stored in the biometric activity data store 350 can be associated with and / or stored according to a particular user or a plurality of users, according to a particular biometric activity category, biometric activity context, according to a particular time, location, etc. In some implementations, the information stored in the biometric activity data store 350 can be associated with and / or stored according to a particular environment (e.g., outdoor, indoor, etc.). In some implementations, the information stored in the biometric activity data store 350 can be associated with and / or stored according to a particular entity that is associated with the performance of the biometric activity (e.g., an entity that monitored the biometric activity, a user that is associated with the biometric activity, an entity to which the biometric activity data is transmitted, etc.). For example, biometric activity data may be stored or retrieved from the biometric activity data store 350 that is relevant to a dialogue between a user and an activity agent for determining aPCT / US24 / 54882 07 November 2024 (07.11.2024) fatigue state or fatigue level of a user, providing content, generating dialogue for communication with a user, or providing a recommendation to a user.

[0063] In some implementations, the content data store 360 can store data associated with content. For example, the content can include images, videos, textual descriptions (e.g., static or commonly used phrases, recommendations, dialogue, etc.), audio recordings, audio spectrograms, etc. In some implementations, the information stored in the content data store 360 can be associated with and / or stored according to a particular user or a plurality of users, according to a particular content category', content genre, content context, time, location, content type, content environment, etc. In some implementations, the information stored in the content data store 360 can be associated with and / or stored according to a particular entity that is associated with the content (e.g., an entity that requests the content to be generated, an entity that is to receive the content, an entity that appears in the content, etc.). For example, machine-learned models described herein can reference or retrieve content from the content data store 360 when determining a fatigue state or fatigue level of a user, generating dialogue, when generating recommendations, etc. The content which is referenced or retrieved from the content data store 360 may be associated with a location or preferences of the user who is to receive the content (e.g., an audio recording, audio spectrogram, image, etc., of the user may be referenced by the machine-learned models described herein to determine a fatigue state and / or generate and recommend action for improving the quality of sleep of the user and / or to increase an energy level of the user).

[0064] Machine-learned model data store 370 can store machine-learned models which can be retrieved and implemented by the server computing system 300 for generating distilled or fine-tuned machine-learned models (e.g., distilled or fine-tuned generative machine-learned models) that, in some implementations, can also be provided to the computing device 100. Machine-learned model data store 370 can also store distilled or fine-tuned machine-learned models (e.g., distilled or fine-tuned generative machine-learned models) which can be retrieved and implemented by the computing device 100. In some implementations, the computing device 100 can retrieve and implement machine-learned models which are large parameter models that have not been fine-tuned or distilled. The machine-learned models (including large parameter models and distilled or fine-tuned models) stored at the machine- learned model data store 370 can include generative machine-learned models respectively associated with different types of applications, types of items, etc., that may be implemented across a variety of domains (e.g., healthcare, gaming, engineering / science, entertainment,PCT / US24 / 54882 07 November 2024 (07.11.2024) travel, retail, etc.). The machine-learned models may include large language models and general, multimodal models (e.g., Gemini). The machine-learned models may include text- to-text large language models, text-to-image large language models, etc. The machine- learned models may include language models which have been trained using reinforcement learning from human feedback. The machine-learned models may include generative artificial intelligence (Al) models which may implement generative adversarial networks (GANs), transformers, variational autoencoders (VAEs), neural radiance fields (NeRFs), and the like.

[0065] External content 500 can be any form of external content including news articles, webpages, image files, video files, audio files, written descriptions, ratings, game content, social media content, photographs, commercial offers, transportation method, weather conditions, sensor data obtained by various sensors, or other suitable external content. The computing device 100, external computing device 200, and server computing system 300 can access external content 500 over network 400. External content 500 can be searched by computing device 100, external computing device 200, and server computing system 300 according to known searching methods and search results can be ranked according to relevance, popularity, or other suitable attributes, including location-specific filtering or promotion.

[0066] FIG. IB is an example block diagram of a computing system, according to one or more example embodiments of the disclosure. Referring now to FIG. IB, example block diagrams of a system 1200 including a computing device 100 and server computing system 300 according to one or more example embodiments of the disclosure will now be described. Although computing device 100 is represented in FIG. IB, features of the computing device 100 described herein are also applicable to the external computing device 200.

[0067] The computing device 100 may include one or more processors 110, one or more memory devices 120, an application system 130, a position determination device 140, an input device 150, a display device 160, an output device 170, a capture device 180, and one or more sensors 190. The server computing system 300 may include one or more processors 310, one or more memory devices 320, and an application system 330.

[0068] For example, the one or more processors 110, 310 can be any suitable processing device that can be included in a computing device 100 or server computing system 300. For example, the one or more processors 110, 310 may include one or more of a processor,PCT / US24 / 54882 07 November 2024 (07.11.2024) processor cores, a controller and an arithmetic logic unit, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an image processor, a microcomputer, a field programmable array, a programmable logic unit, an applicationspecific integrated circuit (ASIC), a microprocessor, a microcontroller, etc., and combinations thereof, including any other device capable of responding to and executing instructions in a defined manner. The one or more processors 110, 310 can be a single processor or a plurality of processors that are operatively connected, for example in parallel.

[0069] The one or more memory devices 120, 320 can include one or more non-transitory computer-readable storage mediums, including a Read Only Memory (ROM), Programmable Read Only Memory (PROM), Erasable Programmable Read Only Memory (EPROM), and flash memory, a USB drive, a volatile memory device including a Random Access Memory (RAM), a hard disk, floppy disks, a Blu-ray disk, or optical media such as CD ROM discs and DVDs, and combinations thereof. However, examples of the one or more memory devices 120, 320 are not limited to the above description, and the one or more memory devices 120, 320 may be realized by other various devices and structures as would be understood by those skilled in the art.

[0070] For example, the one or more memory devices 120 can also include data 122 and instructions 124 that can be retrieved, manipulated, created, or stored by the one or more processors 110. In some example embodiments, such data can be accessed and used as input to implement navigation application 132, and to execute the instructions to perform operations including receiving a voice input from a user describing a biometric state of the user; determining, based on context information extracted from the voice input, first fatigue information associated with the user; determining, based on acoustic-based features extracted from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user, as described according to examples of the disclosure.

[0071] For example, the one or more memory devices 320 can also include data 322 and instructions 324 that can be retrieved, manipulated, created, or stored by the one or more processors 310. In some example embodiments, such data can be accessed and used as input to implement biometric activity application 332, and to execute the instructions to perform operations including receiving a voice input from a user describing a biometric state of the user; determining, based on context information extracted from the voice input, first fatiguePCT / US24 / 54882 07 November 2024 (07.11.2024) information associated with the user; determining, based on acoustic-based features extracted from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user, as described according to examples of the disclosure.

[0072] In some example embodiments, the computing device 100 includes an application system 130. For example, the application system 130 may include the biometric activity application 132. The application system 130 can include various other applications including health applications, content generation applications, search applications, gaming applications, document applications, text messaging applications, email applications, dictation applications, virtual keyboard applications, browser applications, map applications, social media applications, navigation applications, etc.

[0073] According to examples of the disclosure, the biometric activity application 132 may be executed by the computing device 100 to invoke an activity agent that implements one or more machine-learned models to determine a fatigue level or fatigue state of the user and to interact or communicate with a user (e.g., via dialogue operations) regarding biometric activities that a user performs (e.g., exercise, sleep, nutrition, etc ). In some implementations, the activity agent can implement the one or more machine-learned models to provide feedback to the user regarding a biometric activity, provide recommendations to the user, update the user with whether certain biometric goals are being met, ask open-ended questions, etc. In some implementations, the biometric activity application 132 may be part of another application (e.g., a search application, health application, gaming application, etc.) or may be a standalone application. The biometric activity application 132 may be configured to be dynamically interactive according to various user inputs. Example implementations of the biometric activity application 132 are described herein, however the disclosure is not limited to these examples as various modifications may be made to the embodiments described herein.

[0074] In some examples, one or more aspects of the biometric activity application 132 may be implemented by the biometric activity application 332 of the server computing system 300 which may be remotely located, to implement the functions and operations of the activity agent described herein, via one or more machine-learned models. In some examples, one or more aspects of the biometric activity application 332 may be implemented by the biometricPCT / US24 / 54882 07 November 2024 (07.11.2024) activity application 132 of the computing device 100, to implement the functions and operations of the activity agent described herein, via one or more machine-learned models.

[0075] In some example embodiments, the computing device 100 includes a position determination device 140. Position determination device 140 can determine a current geographic location of the computing device 100 and communicate the geographic location to the server computing system 300 over network 400. The position determination device 140 can be any device or circuitry for analyzing the position of the computing device 100. For example, the position determination device 140 can determine actual or relative position by using a satellite navigation positioning sy stem (e.g. a GPS system, a Galileo positioning system, the GLObal Navigation satellite system (GLONASS), the BeiDou Satellite Navigation and Positioning system), an inertial navigation system, a dead reckoning system, based on an IP address, by using triangulation and / or proximity to cellular towers or WiFi hotspots, and / or other suitable techniques for determining a position of the computing device 100. For example, in some implementations the biometric activity application 132 may be configured to utilize position information determined by the position determination device 140 in connection with determining a fatigue level or fatigue state of the user, generating the recommendations or feedback to a user regarding a biometric activity, etc. For example, the biometric activity application 132 may be configured to provide an insight regarding a user’s fatigue level based in part on a distance travelled by the user based on information obtained via the position determination device 140 (e.g., a position of a user may be used by the biometric activity application 132 to determine whether a user is travelling when determining whether a user’s sleep may be affected due to the travel).

[0076] The computing device 100 may include an input device 150 configured to receive an input from a user and may include, for example, one or more of a keyboard (e.g., a physical keyboard, virtual keyboard, etc ), a mouse, a joystick, a button, a switch, an electronic pen or stylus, a gesture recognition sensor (e.g., to recognize gestures of a user including movements of a body part), an input sound device or speech recognition sensor (e.g., a microphone to receive a voice input such as a voice command or a voice query), a track ball, a remote controller, a portable (e.g., a cellular or smart) phone, a tablet PC, a pedal or footswitch, a virtual-reality device, and so on. The input device 150 may also be embodied by a touch- sensitive display having a touchscreen capability, for example. For example, the input device 150 may be configured to receive an input from a user associated with the input device 150 for executing the biometric activity application 132, for providing a voice input to thePCT / US24 / 54882 07 November 2024 (07.11.2024) biometric activity application 132, for providing feedback to the biometric activity application 132, for communicating with other users, for accepting or declining suggestions or recommendations provided by the computing device 100 with respect to a biometric activity, etc.

[0077] The computing device 100 may include a display device 160 which displays information viewable by the user (e.g., a user interface screen). For example, the display device 160 may be a non-touch sensitive display or a touch-sensitive display. The display device 160 may include a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emiting diode (OLED) display, active matrix organic light emiting diode (AMOLED), flexible display, 3D display, a plasma display panel (PDP), a cathode ray tube (CRT) display, and the like, for example. However, the disclosure is not limited to these example displays and may include other types of displays. The display device 160 can be used by the application system 130 provided at the computing device 100 to display information to a user relating to the sleep quality of a user, to display content relating to a recommendation or advice provided to the user, etc. The display device 160 can be configured to provide, for presentation to a user, one or more user interface screens having user interface elements which are selectable by the user for providing information to the biometric activity application 132 (e.g., for providing feedback, for accepting or declining a recommendation, etc.). The display device 160 may be configured to provide feedback or instructions (guidance) to a user regarding a biometric activity.

[0078] The computing device 100 may include an output device 170 to provide an output to the user and may include, for example, one or more of an audio device (e.g., one or more speakers), a haptic device to provide haptic feedback to a user (e.g., a vibration device), a light source (e.g., one or more light sources such as LEDs which provide visual feedback to a user), a thermal feedback system, and the like. For example, the output device 170 may output dialogue operations between the user and the computing device 100. The output device 170 may provide an output including content in response to receiving an input prompt, an output confirming receipt of the input prompt, an output relating to generating advice or a recommendation, an output for providing feedback or instructions (guidance) to a user regarding an item, an output relating to achieving a biometric goal, etc.

[0079] The computing device 100 may include a capture device 180 that is capable of capturing media content, according to various examples of the disclosure. For example, the capture device 180 can include an image capturer 182 (e.g., a camera) which is configured toPCT / US24 / 54882 07 November 2024 (07.11.2024) capture images (e.g., photos, video, and the like). For example, the image capturer 182 can include one or more cameras having an imaging sensor (e.g., a complementary metal-oxide- semiconductor (CMOS) or charge-coupled device (CCD)). For example, the capture device 180 can include a sound capturer 184 (e.g., a microphone) which is configured to capture sound or audio (e.g., an audio recording). The media content captured by the capture device 180 may be transmitted to one or more of the server computing system 300, sensor data store 340, biometric activity data store 350, content data store 360, and machine-learned model data store 370, for example, via network 400. For example, in some implementations, content which is captured by the capture device 180 may be provided as an input to an activity agent, the biometric activity application 132, one or more machine-learned models, etc., for various tasks associated with determining a fatigue level or fatigue state of a user, for determining a recommendation and / or a dialogue operation, etc., as described herein.

[0080] The computing device 100 may include one or more sensors 190. For example, the one or more sensors 190 may include an inertial measurement unit which includes one or more accelerometers and / or one or more gyroscopes. The one or more accelerometers and one or more gyroscopes may be used to capture motion information with respect to the computing device 100. The motion information obtained via the inertial measurement unit may be associated with the user when the computing device 100 is worn or carried by the user. For example, the one or more sensors 190 may include one or more optical sensors (e.g., one or more photoplethysmography (PPG) sensors, one or more electrocardiogram sensors, etc.). The one or more optical sensors may be configured to provide information about a heart rate of the user, heart rate variability (HRV) information, blood oxygen saturation (SpO2) levels, and the like. The one or more sensors 190 may also include other sensors such as a magnetometer, GPS sensor, proximity sensor, Hall effect sensor, galvanic skin sensors, force sensors, temperature sensors, pressure sensors, noise sensors, and the like. For example, in some implementations, content which is captured by the one or more sensors 190 may be provided as an input to an activity agent, the biometric activity application 132, one or more machine-learned models, etc., for various tasks associated with determining a fatigue level or fatigue state of a user, for determining a recommendation and / or a dialogue operation, etc., as described herein. For example, weather conditions (e.g., temperature, wind, precipitation, etc.) measured by various weather sensors of the computing device 100 may be referenced by one or more machine-learned models when generating a recommendation for a user relating to a biometric activity.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0081] In accordance with example embodiments of the disclosure, the server computing system 300 can include one or more processors 310 and one or more memory devices 320 as described herein. The server computing system 300 may also include an application system 330 which is similar to the application system 130 described herein.

[0082] For example, the application system 330 may include the biometric activity application 332 which performs functions similar to those described herein with respect to biometric activity application 132. In some implementations, one or more machine-learned models (e.g., generative machine-learned models, large language models, etc.) associated with the biometric activity application 332 may be configured to determine a fatigue level or fatigue state of a user, and generate summaries, feedback, recommendations, open-ended questions, etc., as described according to examples of the disclosure (e.g., as described with respect to biometric activity application 132). In some implementations, the biometric activity application 332 may be part of another application (e.g., a search application, health application, gaming application, etc.) or may be a standalone application.

[0083] For example, one or more machine-learned models (e.g., generative machine-learned models, large language models, etc.) associated with the application system 330 (e.g., biometric activity application 332) may be configured to perform a first action (e.g., generate a summary of a voice input), while the computing device 100 (e.g., biometric activity application 132) may be configured to perform a second action (e.g., extract audio biomarkers from the voice input, determine a fatigue level or fatigue state of the user, etc.). For example, one or more machine-learned models (e.g., generative machine-learned models, large language models, etc.) associated with the application system 130 (e.g., biometric activity application 132) may be configured to perform a first action (e.g., generate a summary of a voice input), while the server computing system 300 (e.g., biometric activity application 332) may be configured to perform a second action (e.g., extract audio biomarkers from the voice input, determine a fatigue level or fatigue state of the user, etc.).

[0084] Examples of the disclosure are directed to computer implemented methods for application systems which may implement one or more machine-learned models for determining a fatigue level or fatigue state of a user, for generating recommendations, advice, summaries, content, and / or dialogue operations, etc., for example, in association with an activity agent that interacts with the user.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0085] The flow diagrams of FIGS. 2A through 2C illustrate various methods for providing, for determining biometric information associated with a user including a fatigue state or fatigue level of the user. Although shown in a particular sequence or order, unless otherwise specified, the order of the processes can be modified. Thus, the illustrated embodiments should be understood only as examples, and the illustrated processes can be performed in a different order, and some processes can be performed in parallel. Additionally, one or more processes can be omitted in various embodiments. Thus, not all processes are required in every embodiment. Other process flows are possible.

[0086] The operations of FIG. 2A through 2C will be explained with reference to FIG. 3. FIG. 3 illustrates an example block diagram of a system including a biometric activity application, according to one or more example embodiments of the disclosure.

[0087] Referring to FIG. 2A, at operation 2110 the method 2100 includes a computing system receiving a voice input from a user describing a biometric state of the user. As described herein, the computing system may be embodied as computing device 100, for example. The computing system 3100 of FIG. 3 (which can correspond to computing device 100) includes a biometric activity application 3110 having an activity agent 3120 (e.g., a “health coach”). In some implementations, the activity agent 3120 can include components or features including a transcriber 3122, a context determiner 3124, an audio biomarker determiner 3126, a fatigue level determiner 3128, and a recommendation / insight determiner 3129. However, these are example features and the activity agent 3120 may include more or less features. Further, some of these features can be part of the biometric activity application 3110 while not being part of the activity agent 3120. For example, the transcriber 3122 may be omitted and be provided at the server computing system 300 or as a separate feature at the computing device 100.

[0088] Referring to FIG. 3, the voice input 3102 may be provided as an input to the biometric activity application 132, for example, via the input device 150 (e.g., via a microphone). In some implementations, the voice input 3102 may be provided as an input to the biometric activity application 132, for example, via a stored voice recording (e.g., stored in content data store 360, one or more memories 120, one or more memories 320, etc.). For example, the voice input 3102 can indicate or describe a biometric state of the user (e.g., “I am having trouble focusing during meetings,” or “I woke up really tired,” etc.). The voice input 3102 from the user can describe how the user feels, for example, their level of alertness or how the user feels about their sleep or energy state (e.g., at a particular time of day). In somePCT / US24 / 54882 07 November 2024 (07.11.2024) implementations, the voice input 3102 may be sensed or received by the computing device 100 (e.g., the biometric activity application 3110 and / or activity agent 3120) in a passive manner, where the user need not actively request the computing device 100 to process the user’s voice input or the user need not actively provide the input to the computing device 100 (e.g., the biometric activity application 3110 and / or activity agent 3120). In some implementations, the user may launch or execute the biometric activity application 3110 and provide the voice input 3102 to the activity agent 3120 (e.g., via one or more dialogue operations). For example, the user may want to track their health status over time, for example, by keeping a sleep journal or sleep log.

[0089] At operation 2120, the method 2100 includes the computing sy stem generating a transcription of the voice input. In some implementations, the transcriber can be disposed on- device (e.g., at the computing device 100) so that the voice input (or corresponding raw audio or audio data corresponding to the voice input) need not be transmitted to a remote device (e.g., the server computing system 300) for processing, thereby improving the security of the audio data and conserving network resources (e.g., bandwidth resources). For example, in FIG. 3 the activity agent 3120 may include a transcriber 3122 which is configured to generate or obtain the transcription of the voice input 3102. In some implementations, the transcriber 3122 may be configured to label each word utterance with a timestamp (e.g., a millisecond timestamp). In some implementations, the transcriber 3122 may be configured to transcribe the user’s speech voice input in real-time (e.g., when the user provides the voice input 3102 in an active manner). In some implementations, the transcriber 3122 may be configured to transcribe the user’s speech voice input at a later time (e.g., when the user voice input 3102 is collected in a passive manner and the computing device 100 can employ computing resources to perform the task when the computing resources are free or can be scheduled or utilized at an optimal time).

[0090] At operation 2130, the method 2100 includes the computing system obtaining context information from the voice input. For example, in FIG. 3, the context determiner 3124 may be configured to determine the context information which can include or infer one or more of emotional or sentimental state associated with the user or input (e.g., happy, angiy, sad, excited, calm, etc.), a biometric state associated with the user or input (e.g., statements indicative of the user’s health or physical or mental well-being or state of mind), an intent associated with the user or input (e.g., a user’s intent to perform an action, a user’s goal, a user’s desire for information, etc.), temporal context associated with the user or inputPCT / US24 / 54882 07 November 2024 (07.11.2024)(e.g., a scheduling context, a time for a particular action to occur, etc.), a topic associated with the user or input (e.g., a topic, domain, genre, etc., associated with the dialogue), or preference information (e.g., preferences of the user inferred from the dialogue). However, the examples described herein are merely examples and other kinds of context information may be obtained from the voice input (e.g., location or environmental information, a level of urgency, user characteristics, etc.).

[0091] At operation 2140, the method 2100 includes the computing sy stem determining the first fatigue information from the context information. FIG. 2B describes further details regarding how the biometric activity application 3110 (activity agent 3120) can determine the first fatigue information based on context information which is obtained based on the transcription. For example, the first fatigue information can be based on context information which can include one or more of a textual summary generated from the transcription of the voice input, topic information that indicates one or more topics associated with the voice input, sentiment information that indicates one or more sentiments associated with the voice input, etc.

[0092] At operation 2150, the method 2100 includes the computing sy stem obtaining audio biomarkers based on the voice input. For example, in FIG. 3, the audio biomarker determiner 3126 may be configured to obtain the audio biomarkers from the voice input (e.g., raw audio, audio data, etc.) which can indicate fatigue based on changes in vocal features that are affected by cognitive and physical tiredness. For example, fatigue can impact aspects of speech production, such as timing, intensity, and vocal control. The audio biomarkers can include timing-based features that include one or more of a total length of speech, total number of words spoken, rate of speech, speech percentage, mean pause time between words, pause variability, etc. The audio biomarkers can also include a fatigue indication derived from or based on an output of a similarity function that is applied to embeddings associated with a first spectrogram of a current voice input and embeddings associated with a second spectrogram of a baseline voice input. The audio biomarker determiner 3126 can include one or more acoustic machine-learned models trained on acoustic properties such as vocal patterns, tone, intonations, etc., which are configured to extract acoustic-based features (e.g., from spectrograms) for determining whether a user is fatigued, a level of fatigue, and to confirm that the identify of the user.

[0093] At operation 2160, the method 2100 includes the computing sy stem determining the second fatigue information from the audio biomarkers. FIG. 2C describes further detailsPCT / US24 / 54882 07 November 2024 (07.11.2024) regarding how the biometric activity application 3110 (activity agent 3120) can determine the second fatigue information based on the audio biomarkers which are obtained based on the voice input (e.g., raw audio, audio data, etc.). For example, the second fatigue information can be based on audio biomarkers which can include one or more of timing-based features, spectrogram embeddings, fatigue indications or similarity scores obtained based on the spectrograms, etc.

[0094] At operation 2170, the method 2100 includes the computing system implementing one or more first machine-learned models to determine a fatigue level of the user based on the first fatigue information and the second fatigue information. For example, the one or more first machine-learned models may be configured to process, as inputs, the first fatigue information and the second fatigue information, to output a fatigue level or fatigue state associated with the user. In FIG. 3, the biometric activity application 3110 (activity agent 3120) may include a fatigue level determiner 3128 which is configured to determine, via one or more fatigue machine-learned models 3128a, a fatigue output 3150 indicating the fatigue level or fatigue state of the user. For example, the one or more fatigue machine-learned models 3128a may be configured to determine the fatigue level or fatigue state based on the first fatigue information and the second fatigue information.

[0095] In some implementations, the one or more fatigue machine-learned models 3128a may also receive, as a further input, sensor data 3104 and / or biometric activity information 3106 for determining the fatigue level or fatigue state associated with the user. For example, the sensor data can include biometric information of the user, where the biometric information can include ECG information, PPG information, heart rate, heart rate recovery, pulse information, body mass index information, heart rate variability, blood pressure, oxygen saturation, body temperature, sleep metrics (e.g., sleep scores, total sleep time, sleep onset latency, sleep efficiency, sleep stage information, REM sleep latency, number of awakenings, sleep disturbances, etc ), physical activities (e.g., number of steps walked, number of miles cycled, number of laps swam, etc.), audio spectrograms, and the like. The sensor data can also include other information including imagery and / or videos captured by image sensors, lighting information, weather information, noise information, movement or motion information, etc. The biometric activity information 3106 can include information relating to biometric activities of the user. For example, the biometric activity information 3106 can relate to particular biometric activities and can include a time engaged in performing a biometric activity, a metric or score relating to the performance of the biometricPCT / US24 / 54882 07 November 2024 (07.11.2024) activity (e.g., a distance travelled, a number of laps swam, miles bicycled, repetitions and / or sets performed, calories eaten, hours slept, etc.).

[0096] In some implementations, the biometric activity application 3110 may also receive external information 3108 which may include sensor data that is not necessarily biometric information associated with the user, but may include other information (e.g., related to the dialogue and / or biometric activity) such as environmental information (e.g., temperature information, weather information, noise information, calendar information, location information, etc.). The external information 3108 may also external information that is provided via one or more external databases that can be used as part of a retrieval-augmented generation (RAG) framework by the biometric activity application 3110.

[0097] The biometric activity application 3110 (activity agent 3120) can provide an accurate assessment of the fatigue state or alertness level of the user, for example, based on context information associated with the voice input 3102, how a user sounds when they are speaking, and biometric information associated with the user, such as the quality of the sleep of the user (e.g., based on sensor data obtained from a smartphone or wearable sleep tracker). For example, the sensor data 3104 (e.g., sleep staging information), as well as combined audio biomarkers, and context information, can be fed to the one or more fatigue machine-learned models 3128a which can be personalized to the user and can be used to provide an accurate assessment of a biometric state of the user, for example, a fatigue state. In some implementations, the biometric activity application 3110 (activity agent 3120) may be configured to provide a fatigue output 3150 in the form of a score that is easily understandable by the user. In some implementations, the fatigue output 3150 may be in the form of a short descriptive text or summary' that is easily ascertainable by the user. Further, the one or more fatigue machine-learned models 3128a can provide a fatigue output 3150 which corresponds to a sleep need of the user, for example, to reduce a sleep debt of the user, to increase or restore an energy level of the user, etc.

[0098] In some implementations, the one or more outputs of the one or more fatigue machine-learned models 3128a may be provided as an input to the recommendation / insight determiner 3129 which includes one or more recommendation machine-learned models 3129a. For example, the one or more recommendation machine-learned models 3129a may be configured to process the fatigue output 3150 to generate content including a recommendation or insight to reduce a fatigue level associated with the user. For example, based on the fatigue output 3150, the activity agent 3120 can be configured to provide anPCT / US24 / 54882 07 November 2024 (07.11.2024) insight / recommendation output 3160 with insights or recommendations on how a user can improve their sleep (e.g., by providing recommendations on how to fall asleep faster, improve their dietary habits to improve sleep, recommendations to reduce night awakenings, etc.).

[0099] FIG. 4 illustrates an example block diagram of a system including one or more machine-learned models, according to one or more example embodiments of the disclosure. For example, in FIG. 4 the system 4100 includes one or more recommendation machine- learned models 4130 which receives as inputs historical information 4110 and the fatigue output 4120. For example, the biometric activity application 3110 (activity agent 3120) may be configured to retrieve historical information associated with the user. In some implementations, the historical information can a plurality of prior fatigue levels from prior time periods (e.g., prior days) and sleep quality information associated with the prior time periods (e.g., prior nights). In some implementations, the historical information 4110 can include information relating to biometric activities other than sleep. In some implementations, the one or more recommendation machine-learned models 4130 may be configured to process the fatigue output 4120 and the historical information 4110 to generate an output 4140 which can include a recommendation or course of action to reduce a sleep debt or fatigue level associated with the user. For example, the activity agent 3120 can provide the output 4140 to the user via one or more dialogue operations.

[0100] In some implementations, the user can provide feedback 4150 to the activity agent 3120 The biometric activity application 3110 (activity agent 3120) may be configured to analyze the feedback and provide the results of the one or more recommendation machine- learned models 4130 as an input for generating an insight (e.g., a revised insight, a new insight, etc.) or for providing the recommendation (e.g., a revised recommendation, a new recommendation, etc.). The feedback and / or the results of the analysis of the user’s feedback can also be stored (e.g., in machine-learned model data store 370, one or more memories 120, one or more memories 320, etc.). In some implementations, the biometric activity application 3110 including the activity agent 3120 may be configured to retrieve the stored feedback information for generating the insight or recommendation.

[0101] For example, based on the fatigue output 4120 and historical information 4110, the activity agent 3120 can be configured to provide an insight / recommendation output 3160 with insights or recommendations on how a user can reduce their sleep debt or fatigue level (e.g., by adjusting habits, by alerting the user of trends, etc ). For example, based on thePCT / US24 / 54882 07 November 2024 (07.11.2024) fatigue output 4120 (which can be based on prior voice inputs indicating a loss of concentration) and historical information 4110 (which can include a record of cycling biometric activities, a history of hours of sleep each day, etc.), the activity agent 3120 can be configured to provide an insight / recommendation output 3160 recommending to the user to get extra sleep in certain circumstances, such as: “On days you went on long bike rides, your body needs an extra hour of sleep, otherwise you report loss of concentration during the day.” Accordingly, the activity agent 3120 including the one or more fatigue machine-learned models 3128a and / or one or more recommendation machine-learned models 3129a may be configured to provide specific advice to assist the user in minimizing daytime sleepiness, optimize their sleep and / or biometric activities, and help achieve an improved quality of life.

[0102] FIG. 2B is an example method for determining the context information, according to examples of the disclosure. In FIG. 2B, operation 2210 of method 2200 includes generating a transcription of the voice input. Operation 2210 corresponds to operation 2120 of FIG. 2A, and therefore a repeated description thereof will be omitted for the sake of brevity.

[0103] At operation 2220, the method 2200 includes implementing one or more second machine-learned models to generate a summary. For example, in FIG. 3 the context determiner 3124 may include a summary generator 3124a having one or more second machine-learned models (e.g., large language models) which are configured to process the transcription to output a textual summary of the voice input. For example, the one or more second machine-learned models (e.g., large language models) may be configured to generate a summary of the transcription by analyzing the content, parsing the text and identifying or extracting key features (e.g., entities, topics, etc.), condensing the information (e.g., by utilizing key phrases or core ideas, removing redundant information, etc.), and ensuring the final output (e.g., the textual summary) is coherent and concise (e.g., checking grammar, logically organized, etc.). In some implementations, the textual summary may be limited to a predetermined number of words (e.g., no more than 20 words, 50 words, etc.), a predetermined percentage of the original content, etc. The context information can include the textual summary. In some implementations, the textual summary may be stored as part of a sleep log or sleep journal that is associated with the user (e.g., in biometric activity data store 350, one or more memories 120, one or more memories 320, etc.).

[0104] At operation 2230, the method 2200 includes implementing one or more third machine-learned models to extract topic information, sentiment information, etc., from the transcription. For example, in FIG. 3 the context determiner 3124 may include aPCT / US24 / 54882 07 November 2024 (07.11.2024) topic / sentiment classifier 3124b having one or more third machine-learned models (e.g., classifier models such as fine-tuned stand-alone text classification models) which are configured to process the transcription to output one or more labels associated with topics or sentiments in the transcription. In some implementations, the one or more third machine- learned models may be configured to use contextual embeddings, where each token (word) in the transcription can be represented as a vector in a multi-dimensional space. These vectors capture the meaning of the word in the context of surrounding words, allowing the one or more third machine-learned models to understand nuanced meaning and relationships between words. In some implementations, the one or more third machine-learned models may be configured to utilize an attention mechanism to focus or prioritize phrases or sections in the text that determine the topic or sentiment, associate probabilities for each possible label (e.g., possible topic or sentiment), and select one or more predicted labels as having the highest probability(ies) as the predicted topics or sentiments for the transcription. However, other implementation methods may be employed by the one or more machine-learned models for determining the topic information and sentiment information associated with the voice input.

[0105] In some implementations, the one or more third machine-learned models may be implemented in parallel with the one or more second machine-learned models at operation 2220. In some implementations, the one or more third machine-learned models may require less processing power than the one or more second machine-learned models (e.g., the one or more second machine-learned models may utilize billions of parameters while the one or more third machine-learned models may utilize millions of parameters, for example, about ten times less parameters). Further, the smaller model (e.g., the one or more third machine- learned models) can be stored or hosted on a device (e.g., computing device 100) that has less processing power, therefore avoiding the need to transmit certain information to a remote device (e.g., server computing system 300) for performing the operations of the one or more third machine-learned models and thereby enhancing security of the information provided as an input to the one or more third machine-learned models. Accordingly, less computation power may be expended by using a lighter model and disaggregating the tasks using different models, as well as increasing performance speed. As an example, a user may indicate via a voice input that they “slept horribly because the cat kept on crawling on their face” and the one or more third machine-learned models can classify the dialogue information (e.g., with topics such as “trouble with sleep” “cat,” etc.). For example, the context information canPCT / US24 / 54882 07 November 2024 (07.11.2024) include the topics, sentiments, etc., that are identified by the one or more third machine- learned models.

[0106] At operation 2240, the method 2200 includes implementing the one or more first machine-learned models (e.g., from the fatigue level determiner 3128) to determine the fatigue level of the user based on the first fatigue information and the second fatigue information, the first fatigue information being based on at least one of the textual summary (e.g., generated by the summary generator 3124a), the topic information (e.g., determined by the topic / sentiment classifier 3124b), or the sentiment information (e.g., determined by the topic / sentiment classifier 3124b). For example, the first fatigue information can comprise one or more of the textual summan' generated from the transcription of the voice input, the topic information that indicates one or more topics associated with the voice input, and the sentiment information that indicates one or more sentiments associated with the voice input. However, the disclosure is not limited to these examples and the first fatigue information can comprise other contextual information associated with voice input including temporal information, health information, preference information, etc. Operation 2240 is similar to operation 2170 from FIG. 2A, and therefore a repeated description thereof will be omitted for the sake of brevity.

[0107] FIG. 2C is an example method for determining or obtaining audio biomarkers from audio information associated with the voice input. For example, the audio biomarker determiner 3126 may be configured to obtain audio biomarkers from the audio information (e.g., raw audio associated with the voice input). In FIG. 2C, operation 2310 of method 2300 includes receiving a voice input. Operation 2310 corresponds to operation 2110 of FIG. 2A, and therefore a repeated description thereof will be omitted for the sake of brevity.

[0108] At operation 2320, the method 2300 includes generating a transcription of the voice input. Operation 2320 corresponds to operation 2120 of FIG. 2A, and therefore a repeated description thereof will be omitted for the sake of brevity. In some implementations, the transcriber 3122 may be configured to label each word utterance with a timestamp (e.g., a millisecond timestamp). That is, each word utterance in the transcription can be associated with a timestamp. At operation 2330, the method 2300 includes determining timing-based features based on the timestamps. For example, in FIG. 3 the audio biomarker determiner 3126 can include a time-based feature determiner 3126a which is configured to determine, based on the transcription and the timestamp associated with each word utterance in the transcription, a plurality of time-based features. In some implementations, the plurality ofPCT / US24 / 54882 07 November 2024 (07.11.2024) timing-based features can include one or more of a total length of speech, total number of words spoken, rate of speech, speech percentage, mean pause time between words, pause variability, etc. The audio biomarkers can include the plurality of time-based features.

[0109] For example, the timing-based features can be indicative of a user’s cognitive and physical state, including fatigue, through measurable changes in the speaking pattern of the user. For example, a slow rate of speech of the user can be associated with fatigue, as the rate of speech often decreases as the user’s alertness decreases. Similarly, fatigued users tend to have less consistent pauses and thus have a greater pause variability (e.g., a fatigued user may take longer pauses as they think about what to say next or struggle to organize their thoughts, have a lack of concentration, etc.). Speech percentage may correspond to the ratio of speaking time to the total duration of the voice input. Fatigued users may speak for a smaller percentage of time than a non-fatigued person, as a lower speech percentage can reflect reduced engagement and cognitive strain, common indicators of fatigue.

[0110] At operation 2340, the method 2300 includes implementing one or more acoustic machine-learned models to determine acoustic-based features from the voice input. For example, in FIG. 3 the audio biomarker determiner 3126 can include an acoustic-based feature determiner which is configured to process, via one or more machine-learned models (e.g., the one or more acoustic machine-learned models), raw audio associated with the voice input, to determine a fatigue indication. In some implementations, the one or more machine- learned models (e.g., the one or more acoustic machine-learned models) can process the raw audio by extracting a plurality of acoustic-based features from the raw audio associated with the voice input, to determine the fatigue indication. The one or more acoustic machine- learned models can be trained on acoustic properties such as vocal patterns, tone, intonations, etc., and can extract acoustic-based features (e.g., from spectrograms) for determining whether a user is fatigued, a level of fatigue, and to confirm that the identity of the user. In some implementations, the one or more acoustic machine-learned models can be disposed on- device (e.g., at the computing device 100) so that the voice input (or corresponding raw audio or audio data corresponding to the voice input) need not be transmitted to a remote device (e.g., the server computing system 300) for processing, thereby improving the security of the audio data and conserving network resources (e.g., bandwidth resources).

[0111] In some implementations, the acoustic-based feature determiner 3126b may be configured to apply a similarity function to the plurality of acoustic-based features and a plurality of baseline acoustic-based features, to determine or detect changes in a user’s voicePCT / US24 / 54882 07 November 2024 (07.11.2024) that can provide a fatigue indication and / or can be used to confirm the identity of the user.The output of the similarity function may correspond to a similarity score that can be used by the one or more acoustic machine-learned models for outputting a fatigue indication. An example similarity function which comprises the dot product of two voice inputs (e.g., a current voice input Ai and a baseline voice input Bi) is shown below as Equation 1 :

[0112] Other methods for determining a similarity between a current voice input and a baseline voice input can be utilized. For example, measurements including a mean squared error method, a structural similarity index, cosine similarity, etc., can be utilized for determining a similarity metric or value. The acoustic-based feature determiner 3126b may also include one or more machine-learned models (e.g., convolutional neural networks) trained on audio data that can capture complex patterns in spectrograms and can be used to produce embeddings (e.g., feature vectors) for each spectrogram, which can then be compared using methods such as Euclidean distance or cosine similarity methods to determine a similarity metric or value. In some implementations, a baseline may be determined over a plurality of sessions in which a voice input is observed, and the plurality of voice inputs (or more particularly the plurality of spectrograms corresponding to the voice inputs) can be an averaged to derive the baseline. The baseline can also be updated over time as more sessions or interactions with the user are observed. In some implementations, the fatigue indication can be a metric which indicates a first level or a first degree of fatigue (e.g., on a binary scale of fatigued or not fatigued, a score from 0 to 10, etc.) that is indicated by the acoustic-based features. The audio biomarkers can include the plurality of acoustic-based features and / or the fatigue indication. As described herein, the audio biomarkers can be combined with other information such as context information, timing-based features, sensor data, etc., and input to the fatigue level determiner 3128 (comprising one or more machine- learned models) to determine or output an overall fatigue level based on the various information.

[0113] In some implementations, the biometric activity application 3110 may be configured to confirm or verify a determination regarding a level of fatigue of the user which is sensed based on the audio biomarkers determined according to the method 2300. For example, if thePCT / US24 / 54882 07 November 2024 (07.11.2024) timing-based features and / or similarity score indicate the user sounds tired or abnormal, the activity agent 3120 may be configured to output a dialogue operation asking the user if they feel tired and if so, if they can think of a cause. The biometric activity application 3110 (activity agent 3120) may be configured to utilize information from a subsequent response from the user for training or fine-tuning the machine-learned models associated with determining the fatigue state of the user and can also be used for subsequent operations in determining a fatigue level or fatigue state of the user and / or for making recommendations or suggestions to the user.

[0114] FIGS. 5A-5B are example spectrograms of voice inputs, according to one or more example embodiments of the disclosure. The spectrograms may be provided as inputs to the one or more acoustic machine-learned models to determine the acoustic properties and to determine whether a voice input is indicative of a user being fatigued. For example, in FIG. 5 A, the spectrograms 5100 include a first set of spectrograms associated with a first person (person A) 5110, a second set of spectrograms associated with a second person (person B) 5120, and a third set of spectrograms associated with a third person (person C) 5130. In each of the examples, the distance between embeddings from the corresponding spectrograms (e.g., a current spectrogram and a baseline spectrogram) indicates that the user is not fatigued (e.g., a distance of zero or a distance less than a threshold value). For example, in FIG. 5B, the spectrograms 5200 include a first set of spectrograms associated with a first person (person A) 5210, a second set of spectrograms associated with a second person (person B) 5220, and a third set of spectrograms associated with a third person (person C) 5230. In each of the examples, the distance between embeddings from the corresponding spectrograms (e.g., a current spectrogram and a baseline spectrogram) indicates that the user is fatigued (e.g., a distance of one or a distance greater than a threshold value). Methods utilized for determining whether a user is fatigue based on the content of the spectrograms, can also be utilized for identifying the user (confirming the identity of the user) which further ensures that the voice input is compared to a baseline input associated with the same user, for purposes of determining a level or indication of fatigue.

[0115] At operation 2350, the method 2300 includes implementing the one or more first machine-learned models (e.g., from the fatigue level determiner 3128) to determine the fatigue level or fatigue state of the user based on the first fatigue information and the second fatigue information, the second fatigue information being based on at least one of the timingbased features (e.g., determined by the time-based feature determiner 3126a) or the acoustic-PCT / US24 / 54882 07 November 2024 (07.11.2024) based features (e.g., determined by the acoustic-based feature determiner 3126b). For example, the second fatigue information can be based on audio biomarkers which can comprise one or more of the timing-based features (e.g., the total length of speech, total number of words spoken, rate of speech, speech percentage, mean pause time between words, pause variability, etc.), the acoustic-based features (e.g., acoustic properties such as vocal patterns, tone, intonations, etc., a fatigue indication such as a similarity score output from the similarity function, etc.), etc. However, the disclosure is not limited to these examples and the second fatigue information can comprise other audio biomarkers including a pitch variability, snoring patterns, speech fluency, etc., associated with the voice input (e.g., which can be obtained through active or passive methods). Operation 2240 is similar to operation 2170 from FIG. 2A, and therefore a repeated description thereof will be omitted for the sake of brevity.

[0116] FIG. 6 depicts a flowchart of a method 6000 for training one or more machine-learned models according to aspects of the disclosure. For instance, an example machine-learned model can include one or more of a LLM, a generative machine-learned model, etc. For example, the one or more machine-learned models may be configured to implement the operations of the biometric activity applications as described herein.

[0117] FIG. 6 is a flow chart diagram illustrating an example method for training a machine- learned model according to example implementations of aspects of the disclosure. One or more portion(s) of example method 6000 can be implemented by a computing system that includes one or more computing devices such as, for example, computing systems described with reference to the other drawings. Each respective portion of example method 6000 can be performed by any (or any combination) of one or more computing devices. Moreover, one or more portion(s) of example method 6000 can be implemented on the hardware components of the device(s) described herein, for example, to train one or more systems or models. FIG. 6 depicts elements performed in a particular order for purposes of illustration and discussion. Those of ordinary skill in the art, using the disclosures provided herein, will understand that the elements of any of the methods discussed herein can be adapted, rearranged, expanded, omitted, combined, or modified in various ways without deviating from the scope of the disclosure. FIG. 6 is described with reference to elements / terms described with respect to other systems and drawings for exemplary illustrated purposes and is not meant to be limiting. One or more portions of example method 6000 can be performed additionally, or alternatively, by other systems.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0118] At 6002, example method 6000 can include obtaining a training instance. A set of training data can include a plurality of training instances divided between multiple datasets (e.g., a training dataset, a validation dataset, or testing dataset). A training instance can be labeled or unlabeled. Although referred to in example method 6000 as a “training” instance, it is to be understood that runtime inferences can form training instances when a model is trained using an evaluation of the model’s performance on that runtime instance (e.g., online training / leaming). Example datatypes for the training instance and various tasks associated therewith are described throughout the disclosure.

[0119] At 6004, example method 6000 can include processing, using one or more machine- learned models, the training instance to generate an output. The output can be directly obtained from the one or more machine-learned models or can be a downstream result of a chain of processing operations that includes an output of the one or more machine-learned models.

[0120] At 6006, example method 6000 can include receiving an evaluation signal associated with the output. The evaluation signal can be obtained using a loss function. Various determinations of loss can be used, such as mean squared error, likelihood loss, cross entropy loss, hinge loss, contrastive loss, or various other loss functions. The evaluation signal can be computed using known ground-truth labels (e.g., supervised learning), predicted or estimated labels (e.g., semi- or self-supervised learning), or without labels (e.g., unsupervised learning). The evaluation signal can be a reward (e.g., for reinforcement learning). The reward can be computed using a machine-learned reward model configured to generate rewards based on output(s) received. The reward can be computed using feedback data describing human feedback on the output(s).

[0121] At 6008, example method 6000 can include updating the machine-learned model using the evaluation signal. For example, values for parameters of the machine-learned model(s) can be learned, in some embodiments, using various training or learning techniques, such as, for example, backwards propagation. For example, the evaluation signal can be backpropagated from the output (or another source of the evaluation signal) through the machine-learned model(s) to update one or more parameters of the model(s) (e.g., based on a gradient of the evaluation signal with respect to the parameter value(s)). For example, system(s) containing one or more machine-learned models can be trained in an end-to-end manner. Gradient descent techniques can be used to iteratively update the parameters over a number of training iterations. In some implementations, performing backwards propagationPCT / US24 / 54882 07 November 2024 (07.11.2024) of errors can include performing truncated backpropagation through time. Example method 6000 can include implementing a number of generalization techniques (e.g., weight decays, dropouts, etc.) to improve the generalization capability of the models being trained.

[0122] In some implementations, example method 6000 can be implemented for training a machine-learned model from an initialized state to a fully trained state (e.g., when the model exhibits a desired performance profde, such as based on accuracy, precision, recall, etc.).

[0123] In some implementations, example method 6000 can be implemented for particular stages of a training procedure. For instance, in some implementations, example method 6000 can be implemented for pre-training a machine-learned model. Pre-training can include, for instance, large-scale training over potentially noisy data to achieve a broad base of performance levels across a variety of tasks / data types. In some implementations, example method 6000 can be implemented for fine-tuning a machine-learned model. Fine-tuning can include, for instance, smaller-scale training on higher-quality (e.g., labeled, curated, etc.) data. Fine-tuning can affect all or a portion of the parameters of a machine-learned model. For example, various portions of the machine-learned model can be “frozen” for certain training stages. For example, parameters associated with an embedding space can be “frozen” during fine-tuning (e.g., to retain information learned from a broader domain(s) than present in the fine-tuning dataset(s)). An example fine-tuning approach includes reinforcement learning. Reinforcement learning can be based on user feedback on model performance during use.Example Machine-Learned Models

[0124] FIG. 7 is a block diagram of an example processing flow for using machine-learned model(s) 1 to process input(s) 2 to generate output(s) 3.

[0125] Machine-learned model(s) 1 can be or include one or multiple machine-learned models or model components. Example machine-learned models can include neural networks (e.g., deep neural networks). Example machine-learned models can include nonlinear models or linear models. Example machine-learned models can use other architectures in lieu of or in addition to neural networks. Example machine-learned models can include decision tree based models, support vector machines, hidden Markov models, Bayesian networks, linear regression models, k-means clustering models, etc.

[0126] Example neural networks can include feed-forward neural networks, recurrent neural networks (RNNs), including long short-term memory (LSTM) based recurrent neuralPCT / US24 / 54882 07 November 2024 (07.11.2024) networks, convolutional neural networks (CNNs), diffusion models, generative-adversarial networks, or other forms of neural networks. Example neural networks can be deep neural networks. Some example machine-learned models can leverage an attention mechanism such as self-attention. For example, some example machine-learned models can include multiheaded self-attention models.

[0127] Machine-learned model(s) 1 can include a single or multiple instances of the same model configured to operate on data from input(s) 2. Machine-learned model(s) 1 can include an ensemble of different models that can cooperatively interact to process data from input(s) 2. For example, machine-learned model(s) 1 can employ a mixture-of-experts structure. See, e.g., Zhou et al., Mixture-of-Experts with Expert Choice Routing, ARXIV:2202.09368V2 (Oct. 14, 2022).

[0128] Input(s) 2 can generally include or otherwise represent various types of data. Input(s) 2 can include one type or many different types of data. Output(s) 3 can be data of the same type(s) or of different types of data as compared to input(s) 2. Output(s) 3 can include one type or many different types of data.

[0129] Example data types for input(s) 2 or output(s) 3 include natural language text data, software code data (e.g., source code, object code, machine code, or any other form of computer-readable instructions or programming languages), machine code data (e.g., binary code, assembly code, or other forms of machine-readable instructions that can be executed directly by a computer's central processing unit), assembly code data (e.g., low-level programming languages that use symbolic representations of machine code instructions to program a processing unit), genetic data or other chemical or biochemical data, image data, audio data, audiovisual data, haptic data, biometric data, medical data, financial data, statistical data, geographical data, astronomical data, historical data, sensor data generally (e.g., digital or analog values, such as voltage or other absolute or relative level measurement values from a real or artificial input, such as from an audio sensor, light sensor, displacement sensor, etc.), and the like. Data can be raw or processed and can be in any format or schema.

[0130] In multimodal inputs 2 or outputs 3, example combinations of data types include image data and audio data, image data and natural language data, natural language data and software code data, image data and biometric data, sensor data and medical data, etc. It is to be understood that any combination of data types in an input 2 or an output 3 can be present.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0131] An example input 2 can include one or multiple datatypes, such as the example data types noted above. An example output 3 can include one or multiple data types, such as the example data types noted above. The data type(s) of input 2 can be the same as or different from the data type(s) of output 3. It is to be understood that the example data types noted above are provided for illustrative purposes only. Data types contemplated within the scope of the disclosure are not limited to those examples noted above.Example Machine-Learned Sequence Processing Models

[0132] FIG. 8 is a block diagram of an example implementation of an example machine- learned model configured to process sequences of information. For instance, an example implementation of machine-learned model(s) 1 can include machine-learned sequence processing model(s) 4. An example system can pass input(s) 2 to sequence processing model(s) 4. Sequence processing model(s) 4 can include one or more machine-learned components. Sequence processing model(s) 4 can process the data from input(s) 2 to obtain an input sequence 5. Input sequence 5 can include one or more input elements 5-1, 5-2, . . . , 5-M, etc. obtained from input(s) 2. Sequence processing model 4 can process input sequence 5 using prediction layer(s) 6 to generate an output sequence 7. Output sequence 7 can include one or more output elements 7-1, 7-2, . . . , 7-7V, etc. generated based on input sequence 5. The system can generate output(s) 3 based on output sequence 7.

[0133] Sequence processing model(s) 4 can include one or multiple machine-learned model components configured to ingest, generate, or otherwise reason over sequences of information. For example, some example sequence processing models in the text domain are referred to as “Large Language Models,” or LLMs. See, e.g, PaLM 2 Technical Report, GOOGLE, https: / / ai.google / static / documents / palm2techreport.pdf (n.d.). Other example sequence processing models can operate in other domains, such as image domains, see, e.g., Dosovitskiy et al., An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, ARXIV:2010.11929V2 (Jun. 3, 2021), audio domains, see, e.g., Agostinelli et al., MusicIM: Generating Music From Text, ARXIV:2301.11325V1 (Jan. 26, 2023), biochemical domains, see, e.g., Jumper et al., Highly accurate protein structure prediction with AlphaFold, 596 Nature 583 (Aug. 26, 2021), by way of example. Sequence processing model(s) 4 can process one or multiple types of data simultaneously. Sequence processing model(s) 4 can include relatively large models (e.g., more parameters, computationally expensive, etc.), relatively small models (e.g., fewer parameters, computationally lightweight, etc.), or both.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0134] In general sequence processing model(s) 4 can obtain input sequence 5 using data from input(s) 2. For instance, input sequence 5 can include a representation of data from input(s) 2 in a format understood by sequence processing model(s) 4. One or more machine- learned components of sequence processing model(s) 4 can ingest the data from input(s) 2, parse the data into pieces compatible with the processing architectures of sequence processing model(s) 4 (e.g., via “tokenization”), and project the pieces into an input space associated with prediction layer(s) 6 (e.g., via “embedding”).

[0135] Sequence processing model(s) 4 can ingest the data from input(s) 2 and parse the data into a sequence of elements to obtain input sequence 5. For example, a portion of input data from input(s) 2 can be broken down into pieces that collectively represent the content of the portion of the input data. The pieces can provide the elements of the sequence.

[0136] Elements 5-1, 5-2, . . . , 5-M can represent, in some cases, building blocks for capturing or expressing meaningful information in a particular data domain. For instance, the elements can describe “atomic units” across one or more domains. For example, for textual input source(s), the elements can correspond to groups of one or more words or sub-word components, such as sets of one or more characters.

[0137] For example, elements 5-1, 5-2, . . . , 5-M can represent tokens obtained using a tokenizer. For instance, a tokenizer can process a given portion of an input source and output a series of tokens (e.g., corresponding to input elements 5-1, 5-2, . . . , 5-M) that represent the portion of the input source. Various approaches to tokenization can be used. For instance, textual input source(s) can be tokenized using a byte-pair encoding (BPE) technique. See, e.g., Kudo et al., SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing, PROCEEDINGS OF THE 2018 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (System Demonstrations), pages 66-71 (October 31-November 4, 2018), https: / / aclanthology.org / D18-2012.pdf. Imagebased input source(s) can be tokenized by extracting and serializing patches from an image.

[0138] In general, arbitrary data types can be serialized and processed into input sequence 5. It is to be understood that element(s) 5-1, 5-2, . . . , 5-M depicted in FIG. 11 can be the tokens or can be the embedded representations thereof.

[0139] Prediction layer(s) 6 can predict one or more output elements 7-1, 7-2, . . . , 7-Nbased on the input elements. Prediction layer(s) 6 can include one or more machine-learned model architectures, such as one or more layers of learned parameters that manipulate and transformPCT / US24 / 54882 07 November 2024 (07.11.2024) the input(s) to extract higher-order meaning from, and relationships between, input element(s) 5-1, 5-2, . . . , 5-M. In this manner, for instance, example prediction layer(s) 6 can predict new output element(s) in view of the context provided by input sequence 5.

[0140] Prediction layer(s) 6 can evaluate associations between portions of input sequence 5 and a particular output element. These associations can inform a prediction of the likelihood that a particular output follows the input context. For example, consider the textual snippet, “The carpenter’s toolbox was small and heavy. It was full of .” Example prediction layer(s) 6 can identify that “It” refers back to “toolbox” by determining a relationship between the respective embeddings. Example prediction layer(s) 6 can also link “It” to the attributes of the toolbox, such as “small” and “heavy.” Based on these associations, prediction layer(s) 6 can, for instance, assign a higher probability to the word “nails” than to the word “sawdust.”

[0141] A transformer is an example architecture that can be used in prediction layer(s) 4.See, e.g., Vaswani et al., Attention Is All You Need, ASNAN IQG.Q^IGINI (Aug. 2, 2023). A transformer is an example of a machine-learned model architecture that uses an attention mechanism to compute associations between items within a context window. The context window can include a sequence that contains input sequence 5 and potentially one or more output element(s) 7-1, 7-2, . . . , 1-N. A transformer block can include one or more attention layer(s) and one or more post-attention layer(s) (e.g. , feedforward layer(s), such as a multilayer perceptron).

[0142] Prediction layer(s) 6 can include other machine-learned model architectures in addition to or in lieu of transformer-based architectures. For example, recurrent neural networks (RNNs) and long short-term memory (LSTM) models can also be used, as well as convolutional neural networks (CNNs). In general, prediction layer(s) 6 can leverage various kinds of artificial neural networks that can understand or generate sequences of information.

[0143] Output sequence 7 can include or otherwise represent the same or different data types as input sequence 5. For instance, input sequence 5 can represent textual data, and output sequence 7 can represent textual data. Input sequence 5 can represent image, audio, or audiovisual data, and output sequence 7 can represent textual data (e.g., describing the image, audio, or audiovisual data). It is to be understood that prediction layer(s) 6, and any other interstitial model components of sequence processing model(s) 4, can be configured toPCT / US24 / 54882 07 November 2024 (07.11.2024) receive a variety of data types in input sequence(s) 5 and output a variety of data types in output sequence(s) 7.

[0144] Output sequence 7 can have various relationships to input sequence 5. Output sequence 7 can be a continuation of input sequence 5. Output sequence 7 can be complementary to input sequence 5. Output sequence 7 can translate, transform, augment, or otherwise modify input sequence 5. Output sequence 7 can answer, evaluate, confirm, or otherwise respond to input sequence 5. Output sequence 7 can implement (or describe instructions for implementing) an instruction provided via input sequence 5.

[0145] Output sequence 7 can be generated autoregressively. For instance, for some applications, an output of one or more prediction layer(s) 6 can be passed through one or more output layers (e.g., softmax layer) to obtain a probability distribution over an output vocabulary (e.g., a textual or symbolic vocabulary) conditioned on a set of input elements in a context window. In this manner, for instance, output sequence 7 can be autoregressively generated by sampling a likely next output element, adding that element to the context window, and re-generating the probability distribution based on the updated context window, and sampling a likely next output element, and so forth.

[0146] Output sequence 7 can also be generated non-autoregressively. For instance, multiple output elements of output sequence 7 can be predicted together without explicit sequential conditioning on each other. See, e.g., Saharia et al., Non- Autoregressive Machine Translation with Latent Alignments, ARXIV:2004.07437V3 (NOV. 16, 2020).

[0147] Output sequence 7 can include one or multiple portions or elements. In an example content generation configuration, output sequence 7 can include multiple elements corresponding to multiple portions of a generated output sequence (e.g., a textual sentence, values of a discretized waveform, computer code, etc.). In an example classification configuration, output sequence 7 can include a single element associated with a classification output. For instance, an output “vocabulary” can include a set of classes into which an input sequence is to be classified. For instance, a vision transformer block can pass latent state information to a multilayer perceptron that outputs a likely class value associated with an input image.

[0148] FIG. 9 is a block diagram of an example technique for populating an example input sequence 8. Input sequence 8 can include various functional elements that form part of the model infrastructure, such as an element 8-0 obtained from a task indicator 9 that signals toPCT / US24 / 54882 07 November 2024 (07.11.2024) any model (s) that process input sequence 8 that a particular task is being performed (e.g., to help adapt a performance of the model(s) to that particular task). Input sequence 8 can include various data elements from different data modalities. For instance, an input modality 10-1 can include one modality of data. A data-to-sequence model 11-1 can process data from input modality 10-1 to project the data into a format compatible with input sequence 8 (e.g., one or more vectors dimensioned according to the dimensions of input sequence 8) to obtain elements 8-1, 8-2, 8-3. Another input modality 10-2 can include a different modality of data. A data-to-sequence model 11-2 can project data from input modality 10-2 into a format compatible with input sequence 8 to obtain elements 8-4, 8-5, 8-6. Another input modality 10-3 can include yet another different modality of data. A data-to-sequence model 11-3 can project data from input modality 10-3 into a format compatible with input sequence 8 to obtain elements 8-7, 8-8, 8-9.

[0149] Input sequence 8 can be the same as or different from input sequence 5. Input sequence 8 can be a multimodal input sequence that contains elements that represent data from different modalities using a common dimensional representation. For instance, an embedding space can have P dimensions. Input sequence 8 can be configured to contain a plurality of elements that have P dimensions. In this manner, for instance, example implementations can facilitate information extraction and reasoning across diverse data modalities by projecting data into elements in the same embedding space for comparison, combination, or other computations therebetween.

[0150] For example, elements 8-0, . . . , 8-9 can indicate particular locations within a multidimensional embedding space. Some elements can map to a set of discrete locations in the embedding space. For instance, elements that correspond to discrete members of a predetermined vocabulary of tokens can map to discrete locations in the embedding space that are associated with those tokens. Other elements can be continuously distributed across the embedding space. For instance, some data types can be broken down into continuously defined portions (e.g., image patches) that can be described using continuously distributed locations within the embedding space.

[0151] In some implementations, the expressive power of the embedding space may not be limited to meanings associated with any particular set of tokens or other building blocks. For example, a continuous embedding space can encode a spectrum of high-order information. An individual piece of information (e.g., a token) can map to a particular point in that space: for instance, a token for the word “dog” can be projected to an embedded value that points toPCT / US24 / 54882 07 November 2024 (07.11.2024) a particular location in the embedding space associated with canine-related information. Similarly, an image patch of an image of a dog on grass can also be projected into the embedding space. In some implementations, the projection of the image of the dog can be similar to the projection of the word “dog” while also having similarity to a projection of the word “grass,” while potentially being different from both. In some implementations, the projection of the image patch may not exactly align with any single projection of a single word. In some implementations, the projection of the image patch can align with a combination of the projections of the words “dog” and “grass.” In this manner, for instance, a high-order embedding space can encode information that can be independent of data modalities in which the information is expressed.

[0152] Task indicator 9 can include a model or model component configured to identify a task being performed and inject, into input sequence 8, an input value represented by element 8-0 that signals which task is being performed. For instance, the input value can be provided as a data type associated with an input modality and projected along with that input modality (e.g., the input value can be a textual task label that is embedded along with other textual data in the input; the input value can be a pixel-based representation of a task that is embedded along with other image data in the input; etc.). The input value can be provided as a data type that differs from or is at least independent from other input(s). For instance, the input value represented by element 8-0 can be a learned within a continuous embedding space.

[0153] Input modalities 10-1, 10-2, and 10-3 can be associated with various different data types (e.g., as described above with respect to input(s) 2 and output(s) 3).

[0154] Data-to-sequence models 11-1, 11-2, and 11-3 can be the same or different from each other. Data-to-sequence models 11-1, 11-2, and 11-3 can be adapted to each respective input modality 10-1, 10-2, and 10-3. For example, a textual data-to-sequence model can subdivide a portion of input text and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-1, 8-2, 8-3, etc.). An image data-to-sequence model can subdivide an input image and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-4, 8-5, 8-6, etc.). An arbitrary datatype data-to-sequence model can subdivide an input of that arbitrary datatype and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-7, 8-8, 8-9, etc.).

[0155] Data-to-sequence models 11-1, 11-2, and 11-3 can form part of machine-learned sequence processing model(s) 4. Data-to-sequence models 11-1, 11-2, and 11-3 can bePCT / US24 / 54882 07 November 2024 (07.11.2024) jointly trained with or trained independently from machine-learned sequence processing model(s) 4. Data-to-sequence models 11-1, 11-2, and 11-3 can be trained end-to-end with machine-learned sequence processing model(s) 4.Example Machine-Learned Model Development Platform

[0156] FIG. 10 is a block diagram of an example model development platform 12 that can facilitate creation, adaptation, and refinement of example machine-learned models (e.g., machine-learned model(s) 1, sequence processing model(s) 4, etc.). Model development platform 12 can provide a number of different toolkits that developer systems can employ in the development of new or adapted machine-learned models.

[0157] Model development platform 12 can provide one or more model libraries 13 containing building blocks for new models. Model libraries 13 can include one or more pretrained foundational models 13-1, which can provide a backbone of processing power across various tasks. Model libraries 13 can include one or more pre-trained expert models 13-2, which can be focused on performance in particular domains of expertise. Model libraries 13 can include various model primitives 13-3, which can provide low-level architectures or components (optionally pre-trained), which can be assembled in various arrangements as desired.

[0158] Model development platform 12 can receive selections of various model components 14. Model development platform 12 can pass selected model components 14 to a workbench15 that combines selected model components 14 into a development model 16.

[0159] Workbench 15 can facilitate further refinement and adaptation of development model16 by leveraging a number of different toolkits integrated with model development platform 12. For example, workbench 15 can facilitate alignment of the development model 16 with a desired performance profile on various tasks using a model alignment toolkit 17.

[0160] Model alignment toolkit 17 can provide a number of tools for causing development model 16 to generate outputs aligned with desired behavioral characteristics. Alignment can include increasing an accuracy, precision, recall, etc. of model outputs. Alignment can include enforcing output styles, schema, or other preferential characteristics of model outputs. Alignment can be general or domain-specific. For instance, a pre-trained foundational model 13-1 can begin with an initial level of performance across multiple domains. Alignment of the pre-trained foundational model 13-1 can include improving a performance in a particularPCT / US24 / 54882 07 November 2024 (07.11.2024) domain of information or tasks (e.g., even at the expense of performance in another domain of information or tasks).

[0161] Model alignment toolkit 17 can integrate one or more dataset(s) 17-1 for aligning development model 16. Curated dataset(s) 17-1 can include labeled or unlabeled training data. Dataset(s) 17-1 can be obtained from public domain datasets. Dataset(s) 17-1 can be obtained from private datasets associated with one or more developer system(s) for the alignment of bespoke machine-learned model(s) customized for private use-cases.

[0162] Pre-training pipelines 17-2 can include a machine-learned model training workflow configured to update development model 16 over large-scale, potentially noisy datasets. For example, pre-training can leverage unsupervised learning techniques (e.g., de-noising, etc.) to process large numbers of training instances to update model parameters from an initialized state and achieve a desired baseline performance. Pre-training pipelines 17-2 can leverage unlabeled datasets in dataset(s) 17-1 to perform pre-training. Workbench 15 can implement a pre-training pipeline 17-2 to pre-train development model 16.

[0163] Fine-tuning pipelines 17-3 can include a machine-learned model training workflow configured to refine the model parameters of development model 16 with higher-quality data. Fine-tuning pipelines 17-3 can update development model 16 by conducting supervised training with labeled dataset(s) in dataset(s) 17-1. Fine-tuning pipelines 17-3 can update development model 1 by conducting reinforcement learning using reward signals from user feedback signals. Workbench 15 can implement a fine-tuning pipeline 17-3 to fine-tune development model 16.

[0164] Prompt libraries 17-4 can include sets of inputs configured to induce behavior aligned with desired performance criteria. Prompt libraries 17-4 can include few-shot prompts (e.g., inputs providing examples of desired model outputs for prepending to a desired runtime query), chain-of-thought prompts (e.g., inputs providing step-by-step reasoning within the exemplars to facilitate thorough reasoning by the model), and the like.

[0165] Example prompts can be retrieved from an available repository' of prompt libraries 17- 4. Example prompts can be contributed by one or more developer systems using workbench 15.

[0166] In some implementations, pre-trained or fine-tuned models can achieve satisfactory performance without exemplars in the inputs. For instance, zero-shot prompts can includePCT / US24 / 54882 07 November 2024 (07.11.2024) inputs that lack exemplars. Zero-shot prompts can be within a domain within a training dataset or outside of the training domain(s).

[0167] Prompt libraries 17-4 can include one or more prompt engineering tools. Prompt engineering tools can provide workflows for retrieving or learning optimized prompt values. Prompt engineering tools can facilitate directly learning prompt values (e.g., input element values) based one or more training iterations. Workbench 15 can implement prompt engineering tools in development model 16.

[0168] Prompt libraries 17-4 can include pipelines for prompt generation. For example, inputs can be generated using development model 16 itself or other machine-learned models. In this manner, for instance, a first model can process information about a task and output an input for a second model to process in order to perform a step of the task. The second model can be the same as or different from the first model. Workbench 15 can implement prompt generation pipelines in development model 16.

[0169] Prompt libraries 17-4 can include pipelines for context injection. For instance, a performance of development model 16 on a particular task can improve if provided with additional context for performing the task. Prompt libraries 17-4 can include software components configured to identify desired context, retrieve the context from an external source (e.g., a database, a sensor, etc.), and add the context to the input prompt. Workbench 15 can implement context injection pipelines in development model 16.

[0170] Although various training examples described herein with respect to model development platform 12 refer to “pre-training” and “fine-tuning,” it is to be understood that model alignment toolkit 17 can generally support a wide variety of training techniques adapted for training a wide variety of machine-learned models. Example training techniques can correspond to the example training method 6000 described above.

[0171] Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance quality where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution andPCT / US24 / 54882 07 November 2024 (07.11.2024) pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models — e.g., understanding an intent in an unstructured request for a task — while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.

[0172] Model plugin toolkit 18 can include validation tools 18-1. Validation tools 18-1 can include tools that can parse and confirm output(s) of a machine-learned model. Validation tools 18-1 can include engineered heuristics that establish certain thresholds applied to model outputs. For example, validation tools 18-1 can ground the outputs of machine-learned models to structured data sources (e.g., to mitigate “hallucinations”).

[0173] Model plugin toolkit 18 can include tooling packages 18-2 for implementing one or more tools that can include scripts or other executable code that can be executed alongside development model 16. Tooling packages 18-2 can include one or more inputs configured to cause machine-learned model(s) to implement the tools (e.g., few-shot prompts that induce a model to output tool calls in the proper syntax, etc.). Tooling packages 18-2 can include, for instance, fine-tuning training data for training a model to use a tool.

[0174] Model plugin toolkit 18 can include interfaces for calling external application programming interfaces (APIs) 18-3. For instance, in addition to or in lieu of implementing tool calls or tool code directly with development model 16, development model 16 can be aligned to output instruction that initiate API calls to send or obtain data via external systems.

[0175] Model plugin toolkit 18 can integrate with prompt libraries 17-4 to build a catalog of available tools for use with development model 16. For instance, a model can receive, in an input, a catalog of available tools, and the model can generate an output that selects a tool from the available tools and initiates a tool call for using the tool.

[0176] Model development platform 12 can include a computational optimization toolkit 19 for optimizing a computational performance of development model 16. For instance, tools for model compression 19-1 can allow development model 16 to be reduced in size while maintaining a desired level of performance. For instance, model compression 19-1 can include quantization workflows, weight pruning and sparsification techniques, etc. Tools for hardware acceleration 19-2 can facilitate the configuration of the model storage andPCT / US24 / 54882 07 November 2024 (07.11.2024) execution formats to operate optimally on different hardware resources. For instance, hardware acceleration 19-2 can include tools for optimally sharding models for distributed processing over multiple processing units for increased bandwidth, lower unified memory requirements, etc. Tools for distillation 19-3 can provide for the training of lighter- weight models based on the knowledge encoded in development model 16. For instance, development model 16 can be a highly performant, large machine-learned model optimized using model development platform 12. To obtain a lightweight model for running in resource-constrained environments, a smaller model can be a “student model” that learns to imitate development model 1 as a “teacher model.” In this manner, for instance, the investment in learning the parameters and configurations of development model 16 can be efficiently transferred to a smaller model for more efficient inference.

[0177] Workbench 15 can implement one, multiple, or none of the toolkits implemented in model development platform 12. Workbench 15 can output an output model 20 based on development model 1 . Output model 20 can be a deployment version of development model 1 . Output model 20 can be a development or training checkpoint of development model 1 . Output model 20 can be a distilled, compressed, or otherwise optimized version of development model 16.

[0178] FIG. 11 is a block diagram of an example training flow for training a machine-learned development model 16. One or more portion(s) of the example training flow can be implemented by a computing system that includes one or more computing devices such as, for example, computing systems described with reference to the other drawings. Each respective portion of the example training flow can be performed by any (or any combination) of one or more computing devices. Moreover, one or more portion(s) of the example training flow can be implemented on the hardware components of the device(s) described herein, for example, to train one or more systems or models. FIG. 11 depicts elements performed in a particular order for purposes of illustration and discussion. Those of ordinary skill in the art, using the disclosures provided herein, will understand that the elements of any of the methods discussed herein can be adapted, rearranged, expanded, omitted, combined, or modified in various ways without deviating from the scope of the disclosure. FIG. 11 is described with reference to elements / terms described with respect to other systems and drawings for exemplary illustrated purposes and is not meant to be limiting. One or more portions of the example training flow can be performed additionally, or alternatively, by other systems.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0179] Initially, development model 16 can persist in an initial state as an initialized model 21. Development model 16 can be initialized with weight values. Initial weight values can be random or based on an initialization schema. Initial weight values can be based on prior pre-training for the same or for a different model.

[0180] Initialized model 21 can undergo pre-training in a pre-training stage 22. Pre-training stage 22 can be implemented using one or more pre-training pipelines 17-2 over data from dataset(s) 17-1. Pre-training can be omitted, for example, if initialized model 21 is already pre-trained (e.g., development model 16 contains, is, or is based on a pre-trained foundational model or an expert model).

[0181] Pre-trained model 23 can then be a new version of development model 16, which can persist as development model 16 or as a new development model. Pre-trained model 23 can be the initial state if development model 16 was already pre-trained. Pre-trained model 23 can undergo fine-tuning in a fine-tuning stage 24. Fine-tuning stage 24 can be implemented using one or more fine-tuning pipelines 17-3 over data from dataset(s) 17-1. Fine-tuning can be omitted, for example, if a pre-trained model as satisfactory performance, if the model was already fine-tuned, or if other tuning approaches are preferred.

[0182] Fine-tuned model 29 can then be a new version of development model 16, which can persist as development model 16 or as a new development model. Fine-tuned model 29 can be the initial state if development model 16 was already fine-tuned. Fine-tuned model 29 can undergo refinement with user feedback 26. For instance, refinement with user feedback 26 can include reinforcement learning, optionally based on human feedback from human users of fine-tuned model 25. As reinforcement learning can be a form of fine-tuning, it is to be understood that fine-tuning stage 24 can subsume the stage for refining with user feedback 26. Refinement with user feedback 26 can produce a refined model 27. Refined model 27 can be output to downstream system(s) 28 for deployment or further development.

[0183] In some implementations, computational optimization operations can be applied before, during, or after each stage. For instance, initialized model 21 can undergo computational optimization 29-1 (e.g., using computational optimization toolkit 19) before pre-training stage 22. Pre-trained model 23 can undergo computational optimization 29-2 (e.g., using computational optimization toolkit 19) before fine-tuning stage 24. Fine-tuned model 25 can undergo computational optimization 29-3 (e.g., using computational optimization toolkit 19) before refinement with user feedback 26. Refined model 27 canPCT / US24 / 54882 07 November 2024 (07.11.2024) undergo computational optimization 29-4 (e.g., using computational optimization toolkit 19) before output to downstream system(s) 28. Computational optimization(s) 29-1, . . . , 29-4 can all be the same, all be different, or include at least some different optimization techniques.Example Machine-Learned Model Inference System

[0184] FIG. 12 is a block diagram of an inference system for operating one or more machine- learned model(s) 1 to perform inference (e.g., for training, for deployment, etc.). A model host 31 can receive machine-learned model(s) 1. Model host 31 can host one or more model instance(s) 31-1, which can be one or multiple instances of one or multiple models. Model host 31 can host model instance(s) 31-1 using available compute resources 31-2 associated with model host 31.

[0185] Model host 31 can perform inference on behalf of one or more client(s) 32. Client(s) 32 can transmit an input request 33 to model host 31. Using input request 33, model host 31 can obtain input(s) 2 for input to machine-learned model(s) 1. Machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3. Using output(s) 3, model host 31 can return an output payload 34 for responding to input request 33 from client(s) 32. Output payload 34 can include or be based on output(s) 3.

[0186] Model host 31 can leverage various other resources and tools to augment the inference task. For instance, model host 31 can communicate with tool interfaces 35 to facilitate tool use by model instance(s) 31-1. Tool interfaces 35 can include local or remote APIs. Tool interfaces 35 can include integrated scripts or other software functionality.Model host 31 can engage online learning interface(s) 36 to facilitate ongoing improvements to machine-learned model(s) 1. For instance, online learning interface(s) 36 can be used within reinforcement learning loops to retrieve user feedback on inferences served by model host 31. Model host 31 can access runtime data source(s) 37 for augmenting input(s) 2 with additional contextual information. For instance, runtime data source(s) 37 can include a knowledge graph 37-1 that facilitates structured information retrieval for information associated with input request(s) 33 (e.g., a search engine service). Runtime data source(s) 37 can include public or private, external or local database(s) 37-2 that can store information associated with input request(s) 33 for augmenting input(s) 2. Runtime data source(s) 37 can include account data 37-3 which can be retrieved in association with a user account corresponding to a client 32 for customizing the behavior of model host 31 accordingly.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0187] Model host 31 can be implemented by one or multiple computing devices or systems. Client(s) 2 can be implemented by one or multiple computing devices or systems, which can include computing devices or systems shared with model host 31.

[0188] For example, model host 31 can operate on a server system that provides a machinelearning service to client device(s) that operate client(s) 32 (e.g., over a local or wide-area network). Client device(s) can be end-user devices used by individuals. Client device(s) can be server systems that operate client(s) 32 to provide various functionality as a service to downstream end-user devices.

[0189] In some implementations, model host 31 can operate on a same device or system as client(s) 32. Model host 31 can be a machine-learning service that runs on-device to provide machine-learning functionality to one or multiple applications operating on a client device, which can include an application implementing client(s) 32. Model host 31 can be a part of a same application as client(s) 32. For instance, model host 31 can be a subroutine or method implemented by one part of an application, and client(s) 32 can be another subroutine or method that engages model host 31 to perform inference functions within the application. It is to be understood that model host 31 and client(s) 32 can have various different configurations.

[0190] Model instance(s) 31-1 can include one or more machine-learned models that are available for performing inference. Model instance(s) 31-1 can include weights or other model components that are stored in persistent storage, temporarily cached, or loaded into high-speed memory. Model instance(s) 31-1 can include multiple instance(s) of the same model (e.g., for parallel execution of more requests on the same model). Model instance(s) 31-1 can include instance(s) of different model(s). Model instance(s) 31-1 can include cached intermediate states of active or inactive model(s) used to accelerate inference of those models. For instance, an inference session with a particular model may generate significant amounts of computational results that can be re-used for future inference runs (e.g., using a KV cache for transformer-based models). These computational results can be saved in association with that inference session so that session can be executed more efficiently when resumed.

[0191] Compute resource(s) 31-2 can include one or more processors (central processing units, graphical processing units, tensor processing units, machine-learning accelerators, etc.) connected to one or more memory devices. Compute resource(s) 31-2 can include a dynamicPCT / US24 / 54882 07 November 2024 (07.11.2024) pool of available resources shared with other processes. Compute resource(s) 31-2 can include memory devices large enough to fit an entire model instance in a single memory instance. Compute resource(s) 31-2 can also shard model instance(s) across multiple memory devices (e.g., using data parallelization or tensor parallelization, etc.). This can be done to increase parallelization or to execute a large model using multiple memory devices which individually might not be able to fit the entire model into memor .

[0192] Input request 33 can include data for input(s) 2. Model host 31 can process input request 33 to obtain input(s) 2. Input(s) 2 can be obtained directly from input request 33 or can be retrieved using input request 33. Input request 33 can be submitted to model host 31 via an API.

[0193] Model host 31 can perform inference over batches of input requests 33 in parallel. For instance, a model instance 31-1 can be configured with an input structure that has a batch dimension. Separate input(s) 2 can be distributed across the batch dimension (e.g., rows of an array). The separate input(s) 2 can include completely different contexts. The separate input(s) 2 can be multiple inference steps of the same task. The separate input(s) 2 can be staggered in an input structure, such that any given inference cycle can be operating on different portions of the respective input(s) 2. In this manner, for instance, model host 31 can perform inference on the batch in parallel, such that output(s) 3 can also contain the batch dimension and return the inference results for the batched input(s) 2 in parallel. In this manner, for instance, batches of input request(s) 33 can be processed in parallel for higher throughput of output payload(s) 34.

[0194] Output payload 34 can include or be based on output(s) 3 from machine-learned model(s) 1. Model host 31 can process output(s) 3 to obtain output payload 34. This can include chaining multiple rounds of inference (e.g., iteratively, recursively, across the same model(s) or different model(s)) to arrive at a final output for a task to be returned in output payload 34. Output payload 34 can be transmitted to client(s) 32 via an API.

[0195] Online learning interface(s) 36 can facilitate reinforcement learning of machine- learned model(s) 1. Online learning interface(s) 36 can facilitate reinforcement learning with human feedback (RLHF). Online learning interface(s) 36 can facilitate federated learning of machine-learned model(s) 1.

[0196] Model host 31 can execute machine-learned model(s) 1 to perform inference for various tasks using various types of data. For example, various different input(s) 2 andPCT / US24 / 54882 07 November 2024 (07.11.2024) output(s) 3 can be used for various different tasks. In some implementations, input(s) 2 can be or otherwise represent image data. Machine-learned model(s) 1 can process the image data to generate an output. As an example, machine-learned model(s) 1 can process the image data to generate an image recognition output (e.g., a recognition of the image data, a latent embedding of the image data, an encoded representation of the image data, a hash of the image data, etc.). As another example, machine-learned model(s) 1 can process the image data to generate an image segmentation output. As another example, machine-learned model(s) 1 can process the image data to generate an image classification output. As another example, machine-learned model(s) 1 can process the image data to generate an image data modification output (e.g., an alteration of the image data, etc.). As another example, machine-learned model(s) 1 can process the image data to generate an encoded image data output (e.g., an encoded and / or compressed representation of the image data, etc.). As another example, machine-learned model(s) 1 can process the image data to generate an upscaled image data output. As another example, machine-learned model(s) 1 can process the image data to generate a prediction output.

[0197] In some implementations, the task is a computer vision task. In some cases, input(s) 2 includes pixel data for one or more images and the task is an image processing task. For example, the image processing task can be image classification, where the output is a set of scores, each score corresponding to a different object class and representing the likelihood that the one or more images depict an object belonging to the object class. The image processing task may be object detection, where the image processing output identifies one or more regions in the one or more images and, for each region, a likelihood that region depicts an object of interest. As another example, the image processing task can be image segmentation, where the image processing output defines, for each pixel in the one or more images, a respective likelihood for each category in a predetermined set of categories. For example, the set of categories can be foreground and background. As another example, the set of categories can be object classes. As another example, the image processing task can be depth estimation, where the image processing output defines, for each pixel in the one or more images, a respective depth value. As another example, the image processing task can be motion estimation, where the network input includes multiple images, and the image processing output defines, for each pixel of one of the input images, a motion of the scene depicted at the pixel between the images in the network input.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0198] In some implementations, input(s) 2 can be or otherwise represent natural language data. Machine-learned model(s) 1 can process the natural language data to generate an output. As an example, machine-learned model(s) 1 can process the natural language data to generate a language encoding output. As another example, machine-learned model(s) 1 can process the natural language data to generate a latent text embedding output. As another example, machine-learned model(s) 1 can process the natural language data to generate a translation output. As another example, machine-learned model(s) 1 can process the natural language data to generate a classification output. As another example, machine-learned model(s) 1 can process the natural language data to generate a textual segmentation output. As another example, machine-learned model(s) 1 can process the natural language data to generate a semantic intent output. As another example, machine-learned model(s) 1 can process the natural language data to generate an upscaled text or natural language output (e.g., text or natural language data that is higher quality than the input text or natural language, etc.). As another example, machine-learned model (s) 1 can process the natural language data to generate a prediction output (e.g., one or more predicted next portions of natural language content).

[0199] In some implementations, input(s) 2 can be or otherwise represent speech data (e.g., data describing spoken natural language, such as audio data, textual data, etc.). Machine- learned model(s) 1 can process the speech data to generate an output. As an example, machine-learned model(s) 1 can process the speech data to generate a speech recognition output. As another example, machine-learned model(s) 1 can process the speech data to generate a speech translation output. As another example, machine-learned model(s) 1 can process the speech data to generate a latent embedding output. As another example, machine-learned model(s) 1 can process the speech data to generate an encoded speech output (e.g., an encoded and / or compressed representation of the speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate an upscaled speech output (e.g., speech data that is higher quality than the input speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate a textual representation output (e.g., a textual representation of the input speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate a prediction output.

[0200] In some implementations, input(s) 2 can be or otherwise represent latent encoding data (e.g., a latent space representation of an input, etc.). Machine-learned model(s) 1 canPCT / US24 / 54882 07 November 2024 (07.11.2024) process the latent encoding data to generate an output. As an example, machine-learned model(s) 1 can process the latent encoding data to generate a recognition output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a reconstruction output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a search output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a reclustering output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a prediction output.

[0201] In some implementations, input(s) 2 can be or otherwise represent statistical data. Statistical data can be, represent, or otherwise include data computed and / or calculated from some other data source. Machine-learned model(s) 1 can process the statistical data to generate an output. As an example, machine-learned model(s) 1 can process the statistical data to generate a recognition output. As another example, machine-learned model(s) 1 can process the statistical data to generate a prediction output. As another example, machine- learned model(s) 1 can process the statistical data to generate a classification output. As another example, machine-learned model(s) 1 can process the statistical data to generate a segmentation output. As another example, machine-learned model(s) 1 can process the statistical data to generate a visualization output. As another example, machine-learned model(s) 1 can process the statistical data to generate a diagnostic output.

[0202] In some implementations, input(s) 2 can be or otherwise represent sensor data. Machine-learned model(s) 1 can process the sensor data to generate an output. As an example, machine-learned model(s) 1 can process the sensor data to generate a recognition output. As another example, machine-learned model(s) 1 can process the sensor data to generate a prediction output. As another example, machine-learned model(s) 1 can process the sensor data to generate a classification output. As another example, machine-learned model(s) 1 can process the sensor data to generate a segmentation output. As another example, machine-learned model(s) 1 can process the sensor data to generate a visualization output. As another example, machine-learned model(s) 1 can process the sensor data to generate a diagnostic output. As another example, machine-learned model(s) 1 can process the sensor data to generate a detection output.

[0203] In some implementations, machine-learned model(s) 1 can be configured to perform a task that includes encoding input data for reliable and / or efficient transmission or storage (and / or corresponding decoding). For example, the task may be an audio compression task.PCT / US24 / 54882 07 November 2024 (07.11.2024)The input may include audio data and the output may comprise compressed audio data. In another example, the input includes visual data (e.g., one or more images or videos), the output comprises compressed visual data, and the task is a visual data compression task. In another example, the task may comprise generating an embedding for input data (e.g., input audio or visual data). In some cases, the input includes audio data representing a spoken utterance and the task is a speech recognition task. The output may comprise a text output which is mapped to the spoken utterance. In some cases, the task comprises encrypting or decrypting input data. In some cases, the task comprises a microprocessor performance task, such as branch prediction or memory address translation.

[0204] In some implementations, the task is a generative task, and machine-learned model(s) 1 can be configured to output content generated in view of input(s) 2. For instance, input(s) 2 can be or otherwise represent data of one or more modalities that encodes context for generating additional content.

[0205] In some implementations, the task can be a text completion task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent textual data and to generate output(s) 3 that represent additional textual data that completes a textual sequence that includes input(s) 2. For instance, machine-learned model(s) 1 can be configured to generate output(s) 3 to complete a sentence, paragraph, or portion of text that follows from a portion of text represented by input(s) 2.

[0206] In some implementations, the task can be an instruction following task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent instructions to perform a function and to generate output(s) 3 that advance a goal of satisfying the instruction function (e.g., at least a step of a multi-step procedure to perform the function). Output(s) 3 can represent data of the same or of a different modality as input(s) 2. For instance, input(s) 2 can represent textual data (e.g., natural language instructions for a task to be performed) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the instructions (e.g., natural language responses, programming language responses, machine language responses, etc.). Input(s) 2 can represent image data (e.g., image-based instructions for a task to be performed, optionally accompanied by textual instructions) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the instructions (e.g., natural language responses, programming language responses, machine language responses, etc.). One or more output(s) 3 can be iteratively or recursively generated to sequentially processPCT / US24 / 54882 07 November 2024 (07.11.2024) and accomplish steps toward accomplishing the requested functionality. For instance, an initial output can be executed by an external system or be processed by machine-learned model(s) 1 to complete an initial step of performing a function. Multiple steps can be performed, with a final output being obtained that is responsive to the initial instructions.

[0207] In some implementations, the task can be a question answering task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent a question to answer and to generate output(s) 3 that advance a goal of returning an answer to the question (e.g., at least a step of a multi-step procedure to perform the function). Output(s) 3 can represent data of the same or of a different modality as input(s) 2. For instance, input(s) 2 can represent textual data (e.g., natural language instructions for a task to be performed) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the question (e.g., natural language responses, programming language responses, machine language responses, etc.). Input(s) 2 can represent image data (e.g., image-based instructions for a task to be performed, optionally accompanied by textual instructions) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the question (e.g., natural language responses, programming language responses, machine language responses, etc.). One or more output(s) 3 can be iteratively or recursively generated to sequentially process and accomplish steps toward answering the question. For instance, an initial output can be executed by an external system or be processed by machine-learned model(s) 1 to complete an initial step of obtaining an answer to the question (e.g., querying a database, performing a computation, executing a script, etc.). Multiple steps can be performed, with a final output being obtained that is responsive to the question.

[0208] In some implementations, the task can be an image generation task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desired portion of image content. The context can include text data, image data, audio data, etc. Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent image data that depicts imagery related to the context. For instance, machine-learned model(s) 1 can be configured to generate pixel data of an image. Values for channel (s) associated with the pixels in the pixel data can be selected based on the context (e.g., based on a probability determined based on the context).

[0209] In some implementations, the task can be an audio generation task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desiredPCT / US24 / 54882 07 November 2024 (07.11.2024) portion of audio content. The context can include text data, image data, audio data, etc. Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent audio data related to the context. For instance, machine-learned model (s) 1 can be configured to generate waveform data in the form of an image (e.g., a spectrogram). Values for channel(s) associated with pixels of the image can be selected based on the context. Machine-learned model(s) 1 can be configured to generate waveform data in the form of a sequence of discrete samples of a continuous waveform. Values of the sequence can be selected based on the context (e.g., based on a probability determined based on the context).

[0210] In some implementations, the task can be a data generation task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desired portion of data (e.g., data from various data domains, such as sensor data, image data, multimodal data, statistical data, etc.). The desired data can be, for instance, synthetic data for training other machine-learned models. The context can include arbitrary data type(s). Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent data that aligns with the desired data. For instance, machine-learned model(s) 1 can be configured to generate data values for populating a dataset. Values for the data object(s) can be selected based on the context (e.g., based on a probability determined based on the context).Example Computing Systems and Devices

[0211] FIG. 13 is a block diagram of an example networked computing system that can perform aspects of example implementations of the disclosure. The system can include a number of computing devices and systems that are communicatively coupled over a network 49. An example computing device 50 is described to provide an example of a computing device that can perform any aspect of the disclosure (e.g., implementing model host 31, client(s) 32, or both). An example server computing system 60 is described as an example of a server computing system that can perform any aspect of the disclosure (e.g., implementing model host 31, client(s) 32, or both). Computing device 50 and server computing system(s) 60 can cooperatively interact (e.g., over network 49) to perform any aspect of the disclosure (e.g., implementing model host 31, client(s) 32, or both). Model development platform system 70 is an example system that can host or serve model development platform(s) 12 for development of machine-learned models. Third-party system(s) 80 are example system(s) with which any of computing device 50, server computing system(s) 60, or model development platform system(s) 70 can interact in the performance of various aspects of thePCT / US24 / 54882 07 November 2024 (07.11.2024) disclosure (e.g., engaging third-party tools, accessing third-party databases or other resources, etc.).

[0212] Network 49 can be any type of communications network, such as a local area network (e.g., intranet), wide area network (e.g., Internet), or some combination thereof and can include any number of wired or wireless links. In general, communication over network 49 can be carried via any type of wired or wireless connection, using a wide variety of communication protocols (e.g., TCP / IP, HTTP, SMTP, FTP), encodings or formats (e.g., HTML, XML), or protection schemes (e.g., VPN, secure HTTP, SSL). Network 49 can also be implemented via a system bus. For instance, one or more devices or systems of FIG. 13 can be co-located with, contained by, or otherwise integrated into one or more other devices or systems.

[0213] Computing device 50 can be any type of computing device, such as, for example, a personal computing device (e.g., laptop or desktop), a mobile computing device (e.g., smartphone or tablet), a gaming console or controller, a wearable computing device, an embedded computing device, a server computing device, a virtual machine operating on a host device, or any other type of computing device. Computing device 50 can be a client computing device. Computing device 50 can be an end-user computing device. Computing device 50 can be a computing device of a service provided that provides a service to an end user (who may use another computing device to interact with computing device 50).

[0214] Computing device 50 can include one or more processors 51 and a memory 52. Processor(s) 51 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 52 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 52 can store data 53 and instructions 54 which can be executed by processor(s) 51 to cause computing device 50 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein.

[0215] Computing device 50 can also include one or more input components that receive user input. For example, a user input component can be a touch-sensitive component (e.g., a touch-sensitive display screen or a touch pad) that is sensitive to the touch of a user inputPCT / US24 / 54882 07 November 2024 (07.11.2024) object (e.g., a finger or a stylus). The touch-sensitive component can serve to implement a virtual keyboard. Other example user input components include a microphone, camera, LIDAR, a physical keyboard or other buttons, or other means by which a user can provide user input.

[0216] Computing device 50 can store or include one or more machine-learned models 55. Machine-learned models 55 can include one or more machine-learned model(s) 1, such as a sequence processing model 4. Machine-learned models 55 can include one or multiple model instance(s) 31-1. Machine-learned model(s) 55 can be received from server computing system(s) 60, model development platform system 70, third party system(s) 80 (e.g., an application distribution platform), or developed locally on computing device 50. Machine- learned model(s) 55 can be loaded into memory 52 and used or otherwise implemented by processor(s) 51. Computing device 50 can implement multiple parallel instances of machine- learned model(s) 55.

[0217] Server computing system(s) 60 can include one or more processors 61 and a memory 62. Processor(s) 61 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 62 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 62 can store data 63 and instructions 64 which can be executed by processor(s) 61 to cause server computing system(s) 60 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein.

[0218] In some implementations, server computing system 60 includes or is otherwise implemented by one or multiple server computing devices. In instances in which server computing system 60 includes multiple server computing devices, such server computing devices can operate according to sequential computing architectures, parallel computing architectures, or some combination thereof.

[0219] Server computing system 60 can store or otherwise include one or more machine- learned models 65. Machine-learned model (s) 65 can be the same as or different from machine-learned model(s) 55. Machine-learned models 65 can include one or more machine- learned model(s) 1, such as a sequence processing model 4. Machine-learned models 65 canPCT / US24 / 54882 07 November 2024 (07.11.2024) include one or multiple model instance(s) 31-1. Machine-learned model(s) 65 can be received from computing device 50, model development platform system 70, third party system(s) 80, or developed locally on server computing system(s) 60. Machine-learned model(s) 65 can be loaded into memory 62 and used or otherwise implemented by processor(s) 61. Server computing system(s) 60 can implement multiple parallel instances of machine-learned model(s) 65.

[0220] In an example configuration, machine-learned models 65 can be included in or otherwise stored and implemented by server computing system 60 to establish a client-server relationship with computing device 50 for serving model inferences. For instance, server computing system(s) 60 can implement model host 31 on behalf of client(s) 32 on computing device 50. For instance, machine-learned models 65 can be implemented by server computing system 60 as a portion of a web service (e.g., remote machine-learned model hosting service, such as an online interface for performing machine-learned model operations over a network on server computing system(s) 60). For instance, server computing system(s) 60 can communicate with computing device 50 over a local intranet or internet connection. For instance, computing device 50 can be a workstation or endpoint in communication with server computing system(s) 60, with implementation of machine-learned models 65 being managed by server computing system(s) 60 to remotely perform inference (e.g., for runtime or training operations), with output(s) returned (e.g., cast, streamed, etc.) to computing device 50. Machine-learned models 65 can work cooperatively or interoperatively with machine- learned models 55 on computing device 50 to perform various tasks.

[0221] Model development platform system(s) 70 can include one or more processors 71 and a memory 72. Processor(s) 71 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 72 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 72 can store data 73 and instructions 74 which can be executed by processor(s) 71 to cause model development platform system(s) 70 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein. Example operations include the functionality described herein with respect to model development platform 12. This and other functionality can be implemented by developer tool(s) 75.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0222] Third-party system(s) 80 can include one or more processors 81 and a memory 82. Processor(s) 81 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 82 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 82 can store data 83 and instructions 84 which can be executed by processor(s) 81 to cause third-party system(s) 80 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein. Example operations include the functionality described herein with respect to tools and other external resources called when training or performing inference with machine-learned model(s) 1, 4, 16, 20, 55, 65, etc. (e.g., third-party resource(s) 85).

[0223] FIG. 13 illustrates one example arrangement of computing systems that can be used to implement the disclosure. Other computing system configurations can be used as well. For example, in some implementations, one or both of computing system 50 or server computing system(s) 60 can implement all or a portion of the operations of model development platform system 70. For example, computing system 50 or server computing system(s) 60 can implement developer tool(s) 75 (or extensions thereof) to develop, update / train, or refine machine-learned models 1, 4, 16, 20, 55, 65, etc. using one or more techniques described herein with respect to model alignment toolkit 17. In this manner, for instance, computing system 50 or server computing system(s) 60 can develop, update / train, or refine machine- learned models based on local datasets (e.g., for model personalization / customization, as permitted by user data preference selections).

[0224] FIG. 14 is a block diagram of an example computing device 98 that performs according to example embodiments of the disclosure. Computing device 98 can be a user computing device or a server computing device (e.g., computing device 50, server computing system(s) 60, etc.). Computing device 98 can implement model host 31. For instance, computing device 98 can include a number of applications (e.g., applications 1 through N). Each application can contain its own machine learning library and machine-learned model(s). For example, each application can include a machine-learned model. Example applications include a biometric activity application, a navigation application, a content generation application, a text messaging application, an email application, a dictation application, aPCT / US24 / 54882 07 November 2024 (07.11.2024) virtual keyboard application, a browser application, a social media application, a chat application, etc. As illustrated in FIG. 14, each application can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, or additional components. In some implementations, each application can communicate with each device component using an API (e.g., a public API). In some implementations, the API used by each application is specific to that application.

[0225] FIG. 15 is a block diagram of an example computing device 99 that performs according to example embodiments of the disclosure. Computing device 99 can be the same as or different from computing device 98. Computing device 99 can be a user computing device or a server computing device (e.g., computing device 50, server computing system(s) 60, etc.). Computing device 98 can implement model host 31. For instance, computing device 99 can include a number of applications (e.g., applications 1 through N). Each application can be in communication with a central intelligence layer. Example applications include a biometric activity application, a navigation application, a content generation application, a text messaging application, an email application, a dictation application, a virtual keyboard application, a browser application, a social media application, a chat application, etc. In some implementations, each application can communicate with the central intelligence layer (and model (s) stored therein) using an API (e.g., a common API across all applications).

[0226] The central intelligence layer can include a number of machine-learned models. For example, as illustrated in FIG. 15, a respective machine-learned model can be provided for each application and managed by the central intelligence layer. In other implementations, two or more applications can share a single machine-learned model. For example, in some implementations, the central intelligence layer can provide a single model for all of the applications. In some implementations, the central intelligence layer is included within or otherwise implemented by an operating system of computing device 99.

[0227] The central intelligence layer can communicate with a central device data layer. The central device data layer can be a centralized repository of data for computing device 99. As illustrated in FIG. 15, the central device data layer can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, or additional components. In some implementations, thePCT / US24 / 54882 07 November 2024 (07.11.2024) central device data layer can communicate with each device component using an API (e.g., a private API).Additional Disclosure

[0228] The technology discussed herein makes reference to servers, databases, software applications, and other computer-based systems, as well as actions taken and information sent to and from such systems. The inherent flexibility of computer-based systems allows for a great variety of possible configurations, combinations, and divisions of tasks and functionality between and among components. For instance, processes discussed herein can be implemented using a single device or component or multiple devices or components working in combination. Databases and applications can be implemented on a single system or distributed across multiple systems. Distributed components can operate sequentially or in parallel.

[0229] Aspects of the disclosure have been described in terms of illustrative embodiments thereof. Any and all features in the following claims can be combined or rearranged in any way possible, including combinations of claims not explicitly enumerated in combination together, as the example claim dependencies listed herein should not be read as limiting the scope of possible combinations of features disclosed herein. Accordingly, the scope of the disclosure is by way of example rather than by way of limitation, and the subject disclosure does not preclude inclusion of such modifications, variations or additions to the disclosure as would be readily apparent to one of ordinary skill in the art. Moreover, terms are described herein using lists of example elements joined by conjunctions such as “and,” “or,” “but,” etc. It should be understood that such conjunctions are provided for explanatory purposes only. Clauses and other sequences of items joined by a particular conjunction such as “or,” for example, can refer to “and / or,” “at least one of’, “any combination of’ example elements listed therein, etc. Terms such as “based on” should be understood as “based at least in part on.”

[0230] Terms used herein are used to describe the example embodiments and are not intended to limit and I or restrict the disclosure. The singular forms “a,” “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. In this disclosure, terms such as "including", "having", “comprising”, and the like are used to specify features, numbers, steps, operations, elements, components, or combinations thereof,PCT / US24 / 54882 07 November 2024 (07.11.2024) but do not preclude the presence or addition of one or more of the features, numbers, steps, operations, elements, components, or combinations thereof.

[0231] The term "and / or" includes a combination of a plurality of related listed items or any item of the plurality of related listed items. For example, the scope of the expression or phrase "A and / or B" includes the item "A", the item "B", and the combination of items "A and B”.

[0232] In addition, the scope of the expression or phrase "at least one of A or B" is intended to include all of the following: (1) at least one of A, (2) at least one of B, and (3) at least one of A and at least one of B. Likewise, the scope of the expression or phrase "at least one of A, B, or C" is intended to include all of the following: (1) at least one of A, (2) at least one of B, (3) at least one of C, (4) at least one of A and at least one of B, (5) at least one of A and at least one of C, (6) at least one of B and at least one of C, and (7) at least one of A, at least one of B, and at least one of C.

[0233] It will be understood that, although the terms first, second, third, etc., may be used herein to describe various elements, the elements are not limited by these terms. Instead, these terms are used to distinguish one element from another element. For example, without departing from the scope of the disclosure, a first element may be termed as a second element, and a second element may be termed as a first element.

[0234] The term “can” should be understood as referring to a possibility of a feature in various implementations and not as prescribing an ability that is necessarily present in every implementation. For example, the phrase “X can perform Y” should be understood as indicating that, in various implementations, X has the potential to be configured to perform Y, and not as indicating that in every instance X must always be able to perform Y. It should be understood that, in various implementations, X might be unable to perform Y and remain within the scope of the disclosure.

[0235] The term “may” should be understood as referring to a possibility of a feature in various implementations and not as prescribing an ability that is necessarily present in every implementation. For example, the phrase “X may perform Y” should be understood as indicating that, in various implementations, X has the potential to be configured to perform Y, and not as indicating that in every instance X must always be able to perform Y. It should be understood that, in various implementations, X might be unable to perform Y and remain within the scope of the disclosure.PCT / US24 / 54882 07 November 2024 (07.11.2024)

[0236] To the extent terms including "module", and "unit," and the like are used herein, these terms may refer to, but are not limited to, a software or hardware component or device, such as a Field Programmable Gate Array (FPGA) or Application Specific Integrated Circuit (ASIC), which performs certain tasks. A module or unit may be configured to reside on an addressable storage medium and configured to execute on one or more processors. Thus, a module or unit may include, by way of example, components, such as software components, object-oriented software components, class components and task components, processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functionality provided for in the components and modules / units may be combined into fewer components and modules / units or further separated into additional components and modules.

[0237] Aspects of the above-described example embodiments may be recorded in non- transitory computer-readable media including program instructions to implement various operations embodied by a computer. The media may also include, alone or in combination with the program instructions, data files, data structures, and the like. Examples of non- transitory computer-readable media include magnetic media such as hard disks, floppy disks, and magnetic tape; optical media such as CD ROM disks, Blu-Ray disks, and DVDs; magneto-optical media such as optical discs; and other hardware devices that are specially configured to store and perform program instructions, such as semiconductor memory , readonly memory (ROM), random access memory (RAM), flash memory, USB memory, and the like. Examples of program instructions include both machine code, such as produced by a compiler, and files containing higher level code that may be executed by the computer using an interpreter. The program instructions may be executed by one or more processors. The described hardware devices may be configured to act as one or more software modules in order to perform the operations of the above-described embodiments, or vice versa. In addition, a non-transitory computer-readable storage medium may be distributed among computer systems connected through a network and computer-readable codes or program instructions may be stored and executed in a decentralized manner. In addition, the non- transitory computer-readable storage media may also be embodied in at least one application specific integrated circuit (ASIC) or Field Programmable Gate Array (FPGA).

[0238] Each block of the flowchart illustrations may represent a unit, module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternativePCT / US24 / 54882 07 November 2024 (07.11.2024) implementations, the functions noted in the blocks may occur out of order. For example, two blocks shown in succession may in fact be executed substantially concurrently (simultaneously) or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.

[0239] Further to the descriptions above, a user may be provided with controls allowing the user to make an election as to both if and when systems, programs, or features described herein may enable collection of user information (e.g., information about a user’s social network, social actions, or activities, profession, a user’s preferences, or a user’s current location), and if the user is sent content or communications from a server. In addition, certain data may be treated in one or more ways before it is stored or used, so that personally identifiable information is removed. For example, a user’s identity may be treated so that no personally identifiable information can be determined for the user, or a user’s geographic location may be generalized where location information is obtained (such as to a city, ZIP code, or state level), so that a particular location of a user cannot be determined. Thus, the user may have control over what information is collected about the user, how that information is used, and what information is provided to the user.

[0240] While the disclosure has been described with respect to various example embodiments, each example is provided by way of explanation, not limitation of the disclosure. Those skilled in the art, upon attaining an understanding of the foregoing, can readily produce alterations to, variations of, and equivalents to such embodiments. Accordingly, the disclosure does not preclude inclusion of such modifications, variations and / or additions to the disclosed subject matter as would be readily apparent to one of ordinary skill in the art. For example, features illustrated or described as part of one embodiment can be used with another embodiment to yield a still further embodiment. Thus, it is intended that the disclosure covers such alterations, variations, and equivalents.

Claims

PCT / US24 / 54882 07 November 2024 (07.11.2024)WHAT IS CLAIMED IS:

1. A computing system, comprising: one or more memories configured to store instructions; and one or more processors configured to execute the instructions to perform operations, the operations comprising: receiving a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

2. The computing system of claim 1, wherein the operations further comprise obtaining the context information from the voice input, and obtaining the context information from the voice input comprises: obtaining a transcription of the voice input; and processing, by one or more second machine-learned models, the transcription, to output a textual summary of the voice input, and wherein the context information includes the textual summary.

3. The computing system of claim 2, wherein obtaining the context information from the voice input comprises: processing, by one or more third machine-learned models, the transcription, to output at least one of one or more topics or one or more sentiments associated with the voice input, and wherein the context information includes the at least one of the one or more topics or the one or more sentiments.

4. The computing system of claim 1, wherein the operations further comprise obtaining the audio biomarkers from the audio information, and obtaining the audio biomarkers from the audio information comprises:PCT / US24 / 54882 07 November 2024 (07.11.2024) obtaining a transcription of the voice input and associating a timestamp with each word utterance in the transcription; and determining, based on the transcription and the timestamp associated with each word utterance in the transcription, a plurality of time-based features, and wherein the audio biomarkers include the plurality of time-based features.

5. The computing system of claim 4, wherein the plurality of time-based features include at least one of a total length of speech, total number of words spoken, rate of speech, speech percentage, mean pause time between words, or pause variability'.

6. The computing system of claim 1, wherein the operations further comprise obtaining the audio biomarkers from the audio information, and obtaining the audio biomarkers from the audio information comprises: processing, by one or more second machine-learned models, raw audio associated with the voice input, to determine a fatigue indication, and wherein the audio biomarkers include the fatigue indication.

7. The computing system of claim 6, wherein the processing, by the one or more second machine-learned models, the raw audio associated with the voice input, to determine the fatigue indication comprises extracting a plurality of acoustic-based features, and the operations further comprise applying a similarity function to the plurality of acoustic-based features and a plurality of baseline acoustic-based features, to determine the fatigue indication.

8. The computing system of claim 7, wherein the plurality of acoustic-based features related to at least one of vocal patterns, tone, or intonations associated with the voice input of the user.

9. The computing system of claim 1, wherein the operations comprise processing, by one or more second machine-learned models, the fatigue level, to generate content including a recommendation to reduce the fatigue level associated with the user.

10. The computing system of claim 1, wherein the operations compnse:PCT / US24 / 54882 07 November 2024 (07.11.2024) retrieving historical information associated with the user, the historical information including a plurality of prior fatigue levels from prior time periods and sleep quality information associated with the prior time periods; and processing, by one or more second machine-learned models, the fatigue level and the historical information, to generate content including a recommendation to reduce a sleep debt associated with the user.

11. The computing system of claim 1 , wherein the operations further comprise receiving sensor data including biometric information of the user, and processing, by the one or more first machine-learned models, the sensor data, the first fatigue information, and the second fatigue information, to output the fatigue level associated with the user.

12. The computing system of claim 11, wherein the biometric information of the user includes at least one of total sleep time, sleep onset latency, sleep efficiency, sleep stage information, number of awakenings, or sleep disturbances.

13. A computer-implemented method, comprising: receiving, by a computing system comprising one or more processors and one or more first machine-learned models, a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by the one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

14. The computer-implemented method of claim 13, further comprising obtaining the context information from the voice input by: obtaining a transcription of the voice input; and processing, by one or more second machine-learned models, the transcription, to output a textual summary describing the biometnc state of the user, andPCT / US24 / 54882 07 November 2024 (07.11.2024) wherein the context information includes the textual summary.

15. The computer-implemented method of claim 14, further comprising: processing, by one or more third machine-learned models, the transcription, to output at least one of one or more topics or one or more sentiments associated with the voice input, and wherein the context information includes the at least one of the one or more topics or the one or more sentiments.

16. The computer-implemented method of claim 13, further comprising obtaining the audio biomarkers from the audio information by: obtaining a transcription of the voice input and associating a timestamp with each word utterance in the transcription; and determining, based on the transcription and the timestamp associated with each word utterance in the transcription, a plurality of time-based features, and wherein the audio biomarkers include the plurality of time-based features.

17. The computer-implemented method of claim 16, further comprising obtaining the audio biomarkers from the audio information by: processing, by one or more second machine-learned models, raw audio associated with the voice input, to determine a fatigue indication, and wherein the audio biomarkers include the fatigue indication.

18. The computer-implemented method of claim 13, further comprising receiving sensor data including biometric information of the user, the context information includes a textual summary of the voice input, topics associated with the voice input, and sentiments associated with the voice input, the audio biomarkers include time-based features associated with the voice input and acoustic-based features related to at least one of vocal patterns, tone, or intonations associated with the voice input of the user, and processing, by the one or more first machine-learned models, the sensor data, the first fatigue information, and the second fatigue information, to output the fatigue level associated with the user.PCT / US24 / 54882 07 November 2024 (07.11.2024)19. The computer-implemented method of claim 13, further comprising: retrieving historical information associated with the user, the historical information including a plurality of prior fatigue levels from prior time periods and sleep quality information associated with the prior time periods; and processing, by one or more second machine-learned models, the fatigue level and the historical information, to generate content including a recommendation to reduce a sleep debt associated with the user.

20. A non-transitory computer readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations, the operations comprising: receiving a voice input from a user describing a biometric state of the user; determining, based on context information obtained from the voice input, first fatigue information associated with the user; determining, based on audio biomarkers obtained from audio information associated with the voice input, second fatigue information associated with the user; and processing, by one or more first machine-learned models, the first fatigue information and the second fatigue information, to output a fatigue level associated with the user.

Images