Spatial computing and eye-tracking controlled interaction with generative models

Abstract

A spatial computing method is implemented by a computing system comprising a user device. The method comprises receiving user input comprising a user instruction, wherein the user input refers to one or more elements within a field of view of the user device; forming a model input for a generative model system based on the user input; sending the model input to the generative model system and obtaining a corresponding response, and performing an action using the response.

Description

SPATIAL COMPUTING AND EYE-TRACKING CONTROLLED INTERACTION WITH GENERATIVE MODELSPRIORITY CLAIM

[0001] The present application is based on and claims priority to United States Provisional Application Number 63 / 745,707 having a fding date of January 15, 2025. The present application claims priority to and the benefit of each of such applications and incorporates all such applications herein by reference in their entirety.FIELD

[0002] This specification relates to spatial computing and eye-tracking controlled interaction with generative models.BACKGROUND

[0003] Human-computer interaction technologies have evolved significantly since the early days of computing. Starting with basic command-line systems, user interfaces have developed over time to become more visual and interactive. More recently, the field of spatial computing has emerged, enabling users to interact with real, virtual, or mixed reality spaces. Example forms of spatial computing include virtual reality, augmented reality, mixed reality and extended reality.SUMMARY

[0004] Aspects and advantages of embodiments of the present 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.

[0005] In an aspect, the present disclosure provides a first example method which is implemented by a computing system comprising a user device. The first example method includes receiving user input comprising a user instruction, wherein the user input refers to one or more elements within a field of view of the user device. The first example method includes forming a model input for a generative model system based on the user input. The method includes sending the model input to the generative model system and obtaining a corresponding response, and performing an action using the response.

[0006] In an aspect, the present disclosure provides a second example method implemented by one or more computing devices. The second example method comprises monitoring a location of a point of attention of a user within a field of view. The location of the point of attention may be monitored using eye-tracking. The second example method include receiving user input comprising a user instruction, wherein the user input refers to one or more elements within the field of view. The second example method includes forming a model input for a generative model system based on the user input, compnsing determining one or more parts of the model input based on the location of the point of attention at one or more times. The second example method includes sending the model input to the generative model system and obtaining a corresponding response, and performing an action using the response.

[0007] Various example implementations described in this specification relate to spatial computing and eye-tracking controlled interaction with generative models. Rather than navigate various menus and options, a user may convey an instruction directly via user input (e.g. voice input) which refers to one or more elements within the field of view. By forming a model input for a generative model system based on the user input and / or eyetracking data, and using the resulting response to perform an action, various example implementations provide a control scheme which improves compatibility' with the predetermined protocols of the underlying computing hardware. In this way, example implementations described in this specification can expand on the capabilities of existing spatial computing systems by leveraging the functionality' provided by the generative model.

[0008] Other example aspects of the present 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 present disclosure and, together with the description, help explain the related principles.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] So that the subject matter of this specification may be more easily understood, example embodiments will now be described with reference to the accompanying drawings, in which:

[0010] Figure 1A shows the field of view of a wearable device in an example use case scenario according to example implementations of aspects of the present disclosure;

[0011] Figure IB shows the result of earn ing out a user’s instruction in the scenario of Figure 1 A according to example implementations of aspects of the present disclosure;

[0012] Figure 1C illustrates use of eye-tracking to activate I detect an object within a field of view according to example implementations of aspects of the present disclosure;

[0013] Figure 2A and 2B show the field of view of a wearable device in another example use case scenario according to example implementations of aspects of the present disclosure;

[0014] Figure 3A shows the field of view of a wearable device in another example use case scenario according to example implementations of aspects of the present disclosure;

[0015] Figure 3B shows the result of carry ing out a user’s instruction in the scenario of Figure 3 A according to example implementations of aspects of the present disclosure;

[0016] Figure 4A shows the field of view of a wearable device in an example use case scenario according to example implementations of aspects of the present disclosure;

[0017] Figure 4B shows the result of carrying out a user instruction in the scenario of Figure 4A according to example implementations of aspects of the present disclosure;

[0018] Figure 4C shows the result of carry ing out another user instruction in the scenario of Figure 4A;

[0019] Figure 5 A is a flow chart diagram illustrating an example method for performing spatial computing according to example implementations of aspects of the present disclosure;

[0020] Figure 5B is a flow chart diagram illustrating an example method for performing eye-tracking controlled computing according to example implementations of aspects of the present disclosure;Figure 5C is a flow chart diagram illustrating an example method for training a machine-learned model according to example implementations of aspects of the present disclosure;

[0021] Figure 6 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 present disclosure;

[0022] Figure 7 is a block diagram of an example sequence processing model according to example implementations of aspects of the present disclosure;

[0023] Figure 8 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 present disclosure;

[0024] Figure 9 is a block diagram of an example model development platform according to example implementations of aspects of the present disclosure;

[0025] Figure 10 is a block diagram of an example training workflow for training a machine-learned model according to example implementations of aspects of the present disclosure;

[0026] Figure 11 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 present disclosure;

[0027] Figure 12 is a block diagram of an example networked computing system according to example implementations of aspects of the present disclosure;

[0028] Figure 13 is a block diagram of an example computing device according to example implementations of aspects of the present disclosure; and

[0029] Figure 14 is a block diagram of an example computing device according to example implementations of aspects of the present disclosure.DETAILED DESCRIPTION

[0030] In one example, this specification describes a spatial computing method which allows a user to interact with elements of a real or virtual 3D space. The spatial computing method may be implemented by a computing system, which may comprise one or more computing devices at one or more locations.

[0031] In various example implementations, a user input is received which identifies one or more elements within a field of view of a user device. In some examples, the user device is a wearable device such as smart glasses. The wearable device may include one or more cameras which capture the field of view as visual data. In some implementations, the wearable device may comprise a virtual reality or mixed-reality device, e.g. a virtual reality or mixed-reality headset, which may provide the field of view by displaying it to the user. Thus, in various implementations, the field of view may represent 1) a real scene, 2) a mixture of real and virtual elements, or 3) a virtual scene. Moreover, although the field of view may in some cases be displayed to the user, this is not essential. For example, the user device may comprise a smartphone or other device having one or more cameras whichcapture the field of view for processing by the device without necessarily displaying the field of view to the user.

[0032] The one or more elements identified by the user input may, for example, comprise objects (real or virtual) within the field of view. In some examples, the user input may comprise a voice input which identifies elements by way of certain predefined keywords such as “this”, “that” or “here”. The computing system may be configured to detect the keywords in the voice input. To disambiguate the user’s intention, the method may further monitor the location of a point of attention of the user within the field of view, for example using eye-tracking. In particular, the user device may include an eye tracker configured to measure eye positions and eye movement so as to determine the location of the user's point of attention within the field of view. In some examples, the user device may be a wearable device comprising an eye-tracker. However, in other examples, the user device may comprise another type of computing device such as a desktop computing device, which may include an eye tracker to monitor the location of the user’s attention in the field of view defined by the screen of the device. Thus, while some embodiments relate to a spatial computing method in which a user interacts with a real or virtual 3D space, in other embodiments the user interacts with a 2D space defined by the field of view of a display screen.

[0033] In some examples, detection of a predefined keyword (e.g. “this, “that” or “here”) in the voice input may act as a trigger event that identifies that the user’s point of attention is located at a location in the field of view that the user wishes to refer to. That is, the system may temporally correlate the location of the point of attention with the one or more trigger events to determine a location in the field of view that the user wishes to refer to.

[0034] Object detection techniques may be used to analyse the field of view to determine elements represented in the field of view. If the location of the user’s point of attention overlaps with one of these elements when the user speaks a predefined keyword, then the system may determine that the user intends to refer to that element. However, in some scenarios, the visual data for the field of view may be analysed (together with the user input) in other ways, so as to infer that the user wishes to refer to a particular element, without the use of keywords and / or point of attention data. For example, the system may provide the field of view (e.g. in the form of one or more image / video frames) to a vision language model (VLM), and may prompt the VLM with a question asking the VLM whether the user has referred to an element (e.g. object) in the field of view, and if so what it is. In some examples the VLM may return a bounding box for the element. In some cases, theVLM may comprise a remotely hosted large language model. In some examples, the VLM may comprise a lower-power lower latency and / or on device VLM. In examples in which a number of elements / objects are known to the VLM, the question that is included in the prompt may be whether the user referred to a known element / object and if so, which element / object did they refer to.

[0035] The computing system may obtain data relating to the one or more elements. For example, after determining that the user wishes to refer to a detected element within the field of view, the computing system may identify visual data (e g. image and / or video data) for that element. In some examples, the computing system may obtain structured data for the one or more elements, e.g. where the element is represented by one or more markup language elements, or in an object hierarchy. For instance, in some examples, optical character recognition (OCR) and / or layout analysis may be performed on the field of view to determine structured data for written elements shown in the field of view. For example, OCR and / or layout analysis may be performed to generate structured data specifying the location and / or content of areas of the field of view which represent text (e.g. handwritten text).Alternatively, or in addition, structured data may be available for virtual elements within the field of view, e.g. in a mixed-reality or virtual reality scenario.

[0036] The computing system may form a model input for a generative model system based on the user input. The model input may comprise a prompt for the generative model system. In some examples, the model input may include a prompt to generate computer code to carry out a user instruction conveyed by the user input. Alternatively, the model input may include a prompt to generate text or an image based on the user instruction conveyed by the user input. Where the user input comprises a voice input, a transcription of one or more parts of the voice input may be included in the prompt. Alternatively, the voice input may be tokenised, and the tokens may be included in the model input. In some examples, the prompt may include an initial template portion (e.g. “generate code to”) together with a transcription of one or more parts of the user input. Where the prompt is to generate code, it may include instructions to generate the code in any suitable language, e.g. a scripting language such as JavaScript, Python, Perl or a shell scripting language.

[0037] The computing system may also include visual data from the field of view in the model input. For example, the model input may comprise data for the entire visual field of view, optionally together with any available structured data. Alternatively, the model input may include data relating to the one or more elements identified in the user input. Thus, the model input may include image data and / or structured data for the one or more elementsidentified in the user input, in addition to other elements derived from the user input (e.g. transcription elements). In some examples, the portions of the model input may be timestamped with the time of the associated user interaction. For example, a transcription portion may be timestamped based on the starting time of the transcription. Image or structured data for an element identified in the user input may be timestamped based on the time that the user identified the element. In this way, the model input may provide a detailed log which the generative model system may use to infer user intent.

[0038] The generative model system may include any suitable model or models. For example, the generative model system may comprise a trained large language model system and / or a trained image generation model system.

[0039] In examples in which code is generated, the generative model system may include a model which is suitable for generating code based on user instructions, e.g. an appropriate auto-regressive transformer-based neural network model. In some examples, the code that is generated by the generative model system may be validated and / or altered by a dynamic code analyser before it is executed. Suitable techniques for validating and / or altering code to check for and reduce / avoid errors are known per se to those skilled in the art and will not be described here. The spatial computing system may include an appropriate runtime environment configured to execute the finalised code. In various implementations, the runtime environment may for example comprise a Javascript engine such as node.js, a Python interpreter. .NET Common Language Runtime etc.

[0040] The generative model may be a multimodal large language model capable of processing text and / or image and / or audio input. The multimodal large language model may also be capable of generating different ty pes of output, e.g. text or image output. Examples of multimodal large language models include those described in Andreas Steiner et al, ■‘PaliGemma 2: A Family of Versatile VLMs for Transfer”. arXiv preprint arXiv: 2412.03555 (2024), Gemini Team, Google “Gemini 1.5 Unlocking multimodal understanding across millions of tokens of context”. arXiv preprint arXiv: 2403.05530 (2024), Li, J. et al, BLIP -2: Bootstrapping language-image pretraining with frozen image encoders and large language models. arXiv preprint arXiv:2301.12597 (2023); Albayrak, J.B. et al.. Flamingo: a visual language model for few-shot learning. Advances in Neural Information Processing Systems 35, 23716- 23736 (2022); Jia, C. et al, Scaling up visual and vision-language representation learning with noisy text supervision. In: International Conference on Machine Learning, pp.4904-4916. PMLR (2021); OpenAI: GPT-4 technical report. arXiv preprint arXiv:2303.08774 (2023); and Lu, J. et al, Unified-IO: A unified model for vision, language,and multi-modal tasks. arXiv preprint arXiv:2206.08916 (2022). Reference is also directed to: A. Vaswani et al., Attention is all you need, in Advances in neural information processing systems, pages 5998-6008, 2017; J. Hoffmann et al., Training compute-optimal large language models. arXiv preprint arXiv:2203.15556, 2022; Colin Raffel et al., Exploring the limits of transfer learning with a unified text-to-text transformer. arXiv preprintarXiv: 1910.10683, 2019; Daniel Adiwardana et al., Towards a human-like open-domain chatbot. CoRR, abs / 2001.09977, 2020; Tom B Brown et al., Language models are few-shot learners. arXiv preprint arXiv:2005.14165, 2020; J. Austin et al., Program synthesis with large language models. arXiv preprint arXiv:2108.07732, 2021; M. Chen et al., Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374, 2021, and Yujia Li et al. , Competition-level code generation with AlphaCode. Science 378, 1092-1097 (2022).

[0041] In some examples, the generative model system includes an image generation model system, e.g. a model based on a diffusion neural network. One example of a model based on a diffusion neural network is Imagen, further details of which can be found in S. Chitwan, et al. “Photorealistic text-to-image diffusion models with deep language understanding / ’ Advances in neural information processing systems 35 (2022): 36479-36494. Another example of a diffusion neural network is Stable Diffusion, further details of which can found in R. Rombach, et al., “High-Resolution Image Synthesis with Latent Diffusion Models,’’ arXiv:2112.10752 (2021). In some examples, the generative model system may comprise a subject-driven text-to-image generation model. Examples of such systems include Dreambooth, described in “DreamBooth: Fine Tuning Text-to-image Diffusion Models for Subject-Driven Generation”, Nataniel Ruiz et al, arXiv:2208.12242, SuTi, described in “Subject-driven Text-to-image Generation via Apprenticeship Learning”, Wenhu Chen et al, arXiv:2304.00186. and Instruct-Imagen. described in “Instruct-Imagen: Image Generation with Multi-modal Instruction”, Heixiang Hu et al, arXiv:2401.01952).

[0042] Figure 1A shows an example of the field of view of a w earable device (e.g. smart glasses or a mixed reality headset). The wearable device is part of a computing system which includes a number of devices. In this case, the wearable device is paired (e.g., using Bluetooth) to a computing device 102 which includes a display screen, which is shown in the field of view. In some examples, the computing system may communicate with one or more remote computing devices (not shown) configured to deploy a generative model.Alternatively, the generative model may be hosted locally.

[0043] The user provides the following voice input: “I want to create an image based on this cup, with a plant inside it based on the plant on the right”. A model input is formedbased on the user input. The model input may include a text transcription of all of part of the voice input and / or it may include audio data for the voice input. The model input may also include visual data extracted from the field of view, as described above. For example, the model input may comprise one or more frames of the field of view. Alternatively, the model input may comprise image data (e.g. pixel data) for the particular elements referenced in the user command, which may be identified in the field of view using image / video analysis techniques, or eye-tracking / keyword based techniques, as described above. In some cases, the model input may comprise a number of portions (e g. a transcription portion, one or more image / video data portions) which are timestamped with the time of the associated user interaction.

[0044] The finalised model input may comprise a prompt to generate an image based on the user’s instruction. The finalised model input is sent to the generative model system, and the generated image is obtained in response. The system may display the generated image on the display screen of the computing device 102, as shown in Figure IB.

[0045] More generally, as illustrated in the example of Figure 1A and IB, the system performs an action (in this case by displaying the image on the screen 102) using the response provided by the generative model. However, in some examples (e.g. as illustrated in Figure 1 A and IB), the action is performed using a further device which is different to the user device. The further device may be a computing device which is present in the field of the user device. The further device may comprise a display on which the action is performed and / or on which an indication that the action has been performed is provided. Advantageously, by separating the action from the user device in this way, the computational burden is shifted away from the user device (which is not required to perform the action), thus reducing the computational resources required by the user device, enabling lighter and cheaper user devices.

[0046] In some examples, the further device (and / or a display screen of the further device) may be selectively activated / detected when the user looks at it, i.e. when the user’s point of attention, as determined by eye tracking, coincides with the location of the further device in the field of view. This is shown in Figure 1C.

[0047] In some examples, the user instruction may comprise an instruction to move a real-world object shown in the field of view. One or more electromechanical actuators may be configured to move the real -world object in response to an appropriate control signal. In the example of Figure 2A and 2B, the user provides voice input to “move this object there”. The system may identify that the user is referring to the cup 202 (as “this object”) based oneye tracking data, as illustrated by the highlighted area around the cup 202 in Figure 2A. The location 204 corresponding to “there” may also be identified using eye tracking, as illustrated by the highlighted area around the location 204 in Figure 2B. For example, the system may correlate the location of the user’s point of attention, as determined using eye-tracking, with detection of the keywords “this” and “there” in the voice input, so as to identify that the user is referring to the object 202 and to the location 204. The system generates a model input accordingly, and sends the model input to the generative model. In this case, the model input may comprise a prompt to generate computer code to perform the user’s instruction. The resulting computer code may comprise instructions to generate one or more control signals to control one or more mechanical actuators (not shown) to move the object in accordance with the user instruction. In the example of Figure 2A, the mechanical actuator may move the object 202 using a string (not shown) which is attached to a platform that supports the object 202.

[0048] In some examples, the user instruction may comprise an instruction to assign an action to a real-world object shown in the field of view, such that when a particular condition is met, the action is performed. The action may, for example, be to control a real-world device (e g. the brightness of a lamp). In this case, the generative model system may generate computer code which creates a listener (e.g. an event handler) configured to generate appropriate electrical signals for controlling the real-world device when the condition is met. In some cases, the condition may be evaluated by processing image and / or video data from the field of view. As an example, the user may provide the following voice input: “when I rotate this cup, change the brightness of my smart desk lamp”. Other conditions may include the recognition of a gesture and / or of a sound.

[0049] In another example, the user instruction may comprise an instruction to move a real-world object within the field of view when a particular condition is met. One or more electromechanical actuators may be configured to move the real-world object in response to an appropriate control signal. The generative model system may generate computer code which creates a listener (e.g. an event handler) configured to generate appropriate electrical signals for controlling the real-world device when the condition is met. As shown in Figure 3a, the condition may for example be the recognition of a particular hand gesture 302. In some examples, the condition may be the recognition of a particular hand gesture together with the recognition of a particular sound.

[0050] In the example of Figure 3a, the cup 304 is supported on a platform 305 which is movable via an electromechanical actuator (not shown), which is configured to pull a stringto move the platform 305 towards the user. The user may provide the following voice input: “when I make this hand gesture, and when I inhale, make the cup come towards me”. In this case, the generative model system may generate computer code which creates a listener (e.g. an event handler) configured to generate appropriate electrical signals to cause the electromechanical actuator to bring the cup towards the user when the condition is met. Subsequently, the user makes the hand gesture again and inhales, and the cup 304 moves towards the user, as shown in Figure 3b.

[0051] In another example, the instruction may comprise an instruction based on written elements show n in the field of view. As discussed above, OCR and / or layout analysis may be performed to generate structured data specifying the location and / or content of areas of the field of view which represent text (e.g. handwritten text). The system may determine that user input refers to particular text using various techniques described here. Figure 4A illustrates an example in which a user points to handwritten text 402 of the word “London” and provides the voice input “show me a map of this place”. A model input is formed accordingly and sent to the model. The model may generate computer code to cause a computing device (not shown) to control a projector to project an image showing a map of London in the field of view, as shown in Figure 4B. The example may continue, for instance, by the user referring to the handwritten text for “St Pancras Hotel” 404 and “Google 6 Pancras Square” 406, and providing voice input requesting directions from one to the other. An example result is shown in Figure 4C.

[0052] FIG. 5 A is a flowchart illustrating an example method 500 for spatial computing using a generative model, according to some implementations. The method 500 may be performed by a computing system, such as the systems that include a user device as described with reference to FIGS. 1A-4C.

[0053] At operation 502, the method 500 can include receiving user input comprising a user instruction, wherein the user input refers to one or more elements within a field of view of a user device. The user input can be provided in various forms, such as a voice command containing deictic terms (e.g., “this,” “that,” “here”) that reference an object or area a user is observing. The one or more elements can be, for example, physical objects in a real-world environment, virtual objects in an augmented or mixed reality view, or user interface components on a display.

[0054] The user device can be a mixed reality device (e.g., smart glasses), a virtual reality device in which the field of view is synthetic, or another device such as a smartphone or tablet where a camera captures the field of view.

[0055] The field of view can represent sensory information, such as visual data, available to the system at the time the user provides the instruction. In some implementations, the computing system can be distributed, comprising a computing device configured to perform an action that is different from the user device. Such a configuration can allow a lightweight user device to capture the input while a separate computing device (e.g., a laptop, smart display, or desktop computer) performs more computationally intensive tasks.

[0056] The user device and the separate computing device can be communicatively coupled using a wireless protocol (e g., Bluetooth, Wi-Fi, or a cellular network). In some examples, the computing device performing the action can be located within the field of view of the user device. To provide feedback, this computing device can further include a display configured to present the result of the action or an indication that the action has been performed, which can provide a feedback loop for the user.

[0057] At operation 504, the method 500 can include forming a model input for a generative model system based on the user input received at operation 502. In some implementations, this operation can involve programmatically constructing a multimodal input package that can provide the generative model with context about the user’s instruction and the surrounding environment. For instance, the model input can be formulated by combining multiple, concurrent data streams, which may improve the system’s ability to capture the user’s intent.

[0058] In some implementations, forming the model input can include monitoring a location of a point of attention of a user within the field of view, and determining one or more parts of the model input based on the location of the point of attention at one or more times. For example, a user device (e.g., a mixed reality headset, smart glasses, or a smartphone) can be equipped with an integrated eye-tracker. In such examples, the location of the point of attention can be determined using eye tracking.

[0059] The system can be configured to temporally correlate a stream of gaze data from the eye-tracker with specific events, such as keyword triggers detected within a user’s voice command. This correlation can assist the system in disambiguating which element in the field of view a user is referencing when an instruction such as “summarize this article” is provided while the user is looking at a particular document.

[0060] To provide additional context to the generative model, the model input can include visual data corresponding to at least part of the field of view. In some scenarios, for instance, the model input can comprise image data for the entire field of view, which can be captured by a camera on the user device. In other scenarios, the visual data may be moretargeted, and can consist of cropped image data corresponding to a specific element of interest that the user is looking at or has referenced. The visual data can comprise pixel data, which the generative model can be configured to process to analyze visual characteristics of the scene or specific objects within it.

[0061] In addition to visual data, the model input can be augmented with structured data derived from the scene. As one example, the system can be configured to perform Optical Character Recognition (OCR) on the visual data to extract text content and associated layout information (e.g., paragraph structure, headings). This structured information can be included in the model input along with the user’s instruction.

[0062] To provide the generative model with a chronological record, in some examples, each part of the multimodal input, such as a transcription of the voice input or the captured image data, can be individually timestamped before being packaged and sent to the generative model system.

[0063] At operation 506, the method 500 can include sending the model input to the generative model system and obtaining a corresponding response. For instance, a multimodal model input, which can include timestamped visual data, user attention data, and a transcription of a user’s voice input, can be transmitted as a prompt to the generative model system. In some implementations, the generative model system can be executed locally on a user device (e.g., a smartphone, smart watch, wearable device, augmented reality glasses, etc.), or it can be a remotely hosted service accessed via a network interface over a network (e.g., a cellular network, a Wi-Fi network, or a mesh network). The generative model system can be configured to process the model input to generate a response based on the user’s instruction.

[0064] The response can be dependent on an inferred intent of the user instruction. In some implementations, for example, the user instruction can comprise an instruction to generate an image based on one or more elements within the field of view. In such cases, the corresponding response can comprise a generated image. As one example, a user interaction can include a user directing their gaze at a flower and providing a voice input, such as, "’Generate a watercolor painting of this.”

[0065] The generative model system, which can be a text-to-image diffusion model for example, can be configured to process image data of the flower along with the textual instruction to generate a new image file. As described in relation to operation 508, performing an action using the response can comprise displaying the generated image to the user, for example, as a virtual overlay in the user’s field of view.

[0066] In other implementations, such as for functional or automation tasks, obtaining the response can include obtaining computer code. In these scenarios, the response obtained from the generative model can comprise executable or interpretable code. The generative model system utilized for such tasks can be a large language model (LLM) or other sequence processing model as described herein, for instance, an auto-regressive transformer model that has been trained on a corpus of code.

[0067] The model can be configured to generate code in a scripting language (e.g., JavaScript, Python, or a shell script) based on the user’s natural language instruction and the contextual data provided in the model input. As will be described further in relation to operation 508, performing the action using the response can comprise executing the generated computer code to carry out a function corresponding to the user’s instruction.

[0068] At operation 508, the method 500 can include performing an action using the response received from the generative model system. The action performed can be based on the type of response generated. For example, if the user instruction was to generate an image and the response comprises image data, the action can include displaying the generated image on a display (e.g., a screen, a projector, etc.) associated with the user device or another computing device within the user’s field of view.

[0069] In some implementations, the response can comprise executable computer code. When the response from the generative model system comprises computer code, performing the action at operation 508 can include executing the code within a suitable runtime environment. In some examples, the runtime environment can be a JavaScript engine, a Python interpreter, or another sandboxed execution environment managed by the computing system. Executing the computer code can, in some implementations, carry' out the user instruction. For instance, in response to a user instruction to cause a lamp to turn on when the user claps, the generative model system can return code that, when executed, establishes logic to perform the requested action.

[0070] In some examples, the user instruction can include an instruction to perform an action in response to a condition being met. The generated and executed code can, for instance, establish an event handler or a persistent listener configured to monitor system inputs for the specified condition. In this way, the system can be configured to dynamically associate new behaviors with objects or events in the environment.

[0071] As one example, a user instruction can, for example, comprise an instruction to assign an action to a real-world object. The executed code can bind the trigger condition to the resulting action forthat object. The trigger conditions can be multimodal. For example, acondition can include the recognition of a particular gesture (e.g., a specific hand sign, a thumbs-up. or another predefined physical motion) detected by one or more cameras of the computing system. In another example, the condition can include the recognition of a particular sound (e.g., a spoken keyword, a clap, or a snap) detected by one or more microphones of the computing system.

[0072] Upon detection of the specified condition, the event handler can trigger the associated action. The resulting action can fulfill a user instruction to control a real-world object. As an illustration, upon recognizing a clap as the condition, the system can be configured to execute the action of energizing an electromechanical actuator to physically manipulate an object, such as toggling a light switch or opening a smart blind.

[0073] Figure 5B illustrates a detailed flow diagram of an example method 550 for multi-modal spatial computing interaction according to example implementations of the present disclosure. This method leverages distinct streams of input to facilitate intuitive user control over real or virtual environments. The process begins with multi-modal input acquisition, utilizing specific hardware components of the user device, such as smart glasses or a mixed-reality headset equipped with an integrated eye-tracker.

[0074] At step 552, the system continuously monitors the location of the user’s point of attention. This involves resolving the user’s gaze vector relative to the field of view to determine exactly where the user is looking at any given millisecond. Concurrently , at step 554, the system receives user input, which can comprise a voice command captured by a microphone and processed by an onboard speech-to-text engine. The system is configured to listen for and detect predefined keywords within this voice input, such as “this,” “that,” or “here.” These keywords act as trigger events. To disambiguate which element in the potentially cluttered field of view the user is referencing, the system temporally correlates the timestamp of the detected keyword (the trigger event) with the gaze data stream from step 552. By aligning the time the word “this” was spoken with the specific coordinate of the user’s gaze at that moment, the system can isolate the specific object or region of interest.

[0075] Following the acquisition of these inputs, the method proceeds to step 556, where the system performs contextual model input formation. Here, the system intelligently combines the disparate data streams into a single, cohesive model input. This input can be constructed as a structured data package, for example, in a JSON format, containing multiple timestamped components that provide a chronological and contextual record of the user’s interaction. The package may include the text transcription of the user’s voice command alongside specific visual data derived from the gaze selection — such as a cropped imageframe or raw pixel data corresponding solely to the identified element. Furthermore, if the identified element contains text, the system may perform Optical Character Recognition (OCR) to extract this text and include it as structured data within the package. This comprehensive, multi-modal context ensures the generative model receives all necessary information to accurately infer the user’s intent.

[0076] The method then moves to model invocation at step 558. The contextually -rich model input constructed in step 556 is transmitted to a generative model system. This can include sending the data package to a remotely hosted service via a network API call, or passing it to a local model if available. The generative model processes the inputs — visual, textual, and metadata — and returns a corresponding response.

[0077] Finally, at step 560, the system performs an action using the response provided by the generative model. The nature of this action depends on the user’s intent as interpreted by the model. In many example implementations, the response from the generative model comprises executable computer code, such as Python or JavaScript. In such cases, step 560 involves executing this code within a suitable runtime environment provided by the system, such as a Python interpreter or a JavaScript engine. The executed code is then capable of performing complex operations, such as instantiating an event handler. For example, the code might configure the system to monitor for a specific future condition, such as a particular hand gesture or a specific sound, and upon detecting that condition, trigger a physical output, such as sending a control signal to an electromechanical actuator to move a physical object.

[0078] Figure 5C depicts a flowchart of a method 580 for training one or more machine-learned models according to aspects of the present disclosure. One or more portion(s) of example method 580 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 figures. Each respective portion of example method 580 can be performed by any (or any combination) of one or more computing devices. Moreover, one or more portion(s) of example method 580 can be implemented on the hardware components of the device(s) described herein, for example, to train one or more systems or models. Figure 5C 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 present disclosure. Figure 5C is described with reference to elements / terms described with respect to other systems and figures for exemplary illustratedpurposes and is not meant to be limiting. One or more portions of example method 580 can be performed additionally, or alternatively, by other systems.

[0079] At 582, example method 580 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 580 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 data types for the training instance and various tasks associated therewith are described throughout the present disclosure.

[0080] At 584, example method 580 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.

[0081] At 586, example method 580 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).

[0082] At 588, example method 580 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 anumber of training iterations. In some implementations, performing backwards propagation of errors can include performing truncated backpropagation through time. Example method 580 can include implementing a number of generalization techniques (e.g., weight decays, dropouts, etc.) to improve the generalization capability of the models being trained.

[0083] In some implementations, example method 580 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 profile, such as based on accuracy, precision, recall, etc.).

[0084] In some implementations, example method 580 can be implemented for particular stages of a training procedure. For instance, in some implementations, example method 580 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.

[0085] In some implementations, example method 580 can be implemented for finetuning 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., Io retain information learned from a broader domain(s) than present in the fine-tuning dataset(s)). In some implementations, example method 580 uses adapter modules. Adapters can be small trainable layers that are inserted between pre-existing layers of a pre-trained model. During the finetuning process, the original parameters of the pre-trained model are typically frozen, and only the parameters of the adapters are updated.

[0086] In some implementations, example method 580 can be implemented to execute parameter-efficient fine-tuning methods, such as Layerwise Optimization of Residuals (LoRA). LoRA can refine pre-trained models with minimal adjustments to the original parameters. This can be achieved by introducing trainable low-rank matrices that modify the behavior of the pre-trained weights without directly altering them. In some implementations, during fine-tuning, only these auxiliary matrices are updated, which significantly reduces the number of parameters that are trained.

[0087] An example fine-tuning approach includes reinforcement learning.Reinforcement learning can be based on user feedback on model performance during use.

[0088] Figure 6 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.

[0089] 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.

[0090] Machine-learned model(s) 1 can be or include, or otherwise be representative of any one or more of the machine-learned models described above with respect to the preceding figures. For example, machine-learned model(s) 1 can be or include, or otherwise be representative of any one or more of any of the models described herein, etc. Although various features, variations, and implementations described below are described with respect to machine-learned model(s) 1, it is to be understood that such features, variations, and implementations are to be understood as described with respect to each of any of the models described herein, etc., any other machine-learned component described herein.

[0091] Example neural networks can include feed-forward neural networks, recurrent neural networks (RNNs), including long short-term memory (LSTM) based recurrent neural 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.

[0092] 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 multiple different models or multiple different model portions configured to operate on data from input(s) 2.

[0093] 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, a model ensemble can include multiple models that have different attributes (e.g., different architectures, trained with different recipes, etc.). The ensemble can output an overall output based on the individual outputs of the constituent models. In this manner, for instance, the diverse constituent models can work together to provide system-level robustness by effectivelyaggregating over individual strengths and weaknesses of any given model. The respective individual outputs can be combined in a weighted combination, using a voting or routing mechanism, or a learned output layer (e.g., one or more feedforward or fully -connected layers).

[0094] 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). For example, different portions of a model can leam (explicitly or implicitly) different expertise areas, with pathways through the model being selected by a learned routing mechanism that engages the appropriate expert for a given input (e.g., a given portion of an input, such as on a per-token basis). For example, a feedforward network can be sparsely activated for a given portion of an input based on an output of a routing mechanism that processes the portion of the input. In this manner, for instance, the group of activated weights can form an “expert” that is selected by the router. On each forward pass, only a subset of the total model weights may be engaged, thereby decreasing a quantity of operations performed for processing a given input compared to a densely activated model. In this manner, for instance, the expressive and interpretive power of a high-parameter-count model can be achieved with more compute-efficient forward passes.

[0095] 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.

[0096] 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.

[0097] 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.

[0098] An example input 2 can include one or multiple data types, 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 t pe(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 present disclosure are not limited to those examples noted above.

[0099] Figure 7 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-N, etc. generated based on input sequence 5. The system can generate output(s) 3 based on output sequence 7.

[0100] 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 are referred to as language models and can leverage language-based understandings across one or multiple modalities of input information. Sequence processing model(s) 4 can include relatively large models (e.g., more parameters, computationally expensive, etc.), which may be referred to as "Large Language Models” or LLMs. Sequence processing model(s) 4 can include relatively small models (e g., fewer parameters, computationally lightweight, etc.), which may be referred to as “Small Language Models” or SLMs. Example language models include, for instance, models described in Gemma: Open Models Based on Gemini Research and Technology, GOOG E, https: / / arxiv.org / abs / 2403.08295; Gemma 2: Improving Open Language Models at a Practical Size, GOOGLE, https: / / arxiv.org / abs / 2408.00118.

[0101] Sequence processing model(s) 4 can process one or multiple ty pes of data simultaneously. Variations of language models that can perform joint vision and language tasks may be referred to as “Vision-Language Models,” or VLMs. Example VLMs include models described in PaliGemma: A versatile 3B VLM for transfer, GOOGLE, https: / / arxiv.org / abs / 2407.07726; PaliGemma 2: A Family of Versatile VLMs for Transfer, GOOGLE, https: / / arxiv.org / abs / 2412.03555; Flamingo: a Visual Language Model for FewShot Learning, GOOGLE, https: / / arxiv.org / abs / 2204.14198; PaLI: A Jointly-Scaled Multilingual Language-Image Model, GOOGLE, https: / / arxiv.org / abs / 2209.06794.

[0102] Sequence processing model(s) 4 can be multimodal. Example multimodal sequence processing models include, for instance, models described in Gemini: A Family of Highly Capable Multimodal Models, GOOGLE, https: / / arxiv.org / abs / 2312.11805; Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context, GOOGLE, https: / / arxiv.org / abs / 2403.05530.

[0103] Other example sequence processing models can operate to generate outputs or receive inputs in specific 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 <\\ . 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.

[0014] 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”).

[0105] 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.

[0106] Elements 5-1, 5-2, . . . , S-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 textualinput 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.

[0107] 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 -AT) 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. Image-based input source(s) can be tokenized by extracting and serializing patches from an image.

[0108] 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 Figure 7 can be the tokens or can be the embedded representations thereof.

[0109] Prediction layer(s) 6 can predict one or more output elements 7-1, 7-2, . . . , 7-N based 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 transform 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.

[0110] 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.”

[0111] 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, ARXlV:1706.03762v7 (Aug. 2, 2023). A transformer is an example of a machine-learned model architecture that uses anattention mechanism to compute associations betw een 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, . . . , 7-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 multi-layer perceptron).

[0112] 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.

[0113] 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 to receive a variety of data types in input sequence(s) 5 and output a variety of data ty pes in output sequence(s) 7.

[0114] 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.

[0115] 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.

[0116] Output sequence 7 can also be generated non-autoregressively. For instance, multiple output elements of output sequence 7 can be predicted together without explicitsequential conditioning on each other. See, e.g., Saharia et al., Non-Autoregressive Machine Translation with Latent Alignments. ARXIV:2004.07437V3 (NOV. 16, 2020).

[0117] 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.

[0118] Figure 8 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 to 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 modality7of 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 modality7of 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.

[0119] 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.

[0120] 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 datatypes can be broken down into continuously defined portions (e.g., image patches) that can be described using continuously distributed locations within the embedding space.

[0121] 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 to 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.

[0122] 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 learned within a continuous embedding space.

[0123] 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).

[0124] 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 data type and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-7. 8-8, 8-9, etc.).

[0125] 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 be 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.

[0126] Figure 9 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.

[0127] 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. Model primitives 13-3 can include a library of pre-trained adapters or LoRA modules that can adapt a baseline foundational model to align its outputs with a desired performance profile, augment model capabilities (e.g., to adapt to a different input modality, etc.), and the like.

[0128] Model development platform 12 can receive selections of various model components 14. Model development platform 12 can pass selected model components 14 to a workbench 15 that combines selected model components 14 into a development model 16.

[0129] Workbench 15 can facilitate further refinement and adaptation of development model 16 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.

[0130] 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 the 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 particular domain of information or tasks (e.g., even at the expense of performance in another domain of information or tasks).

[0131] 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.

[0132] 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., denoising, 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.

[0133] Fine-tuning pipelines 17-3 can include a machine-learned model training w orkflow 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 16 by conducting reinforcement learning using reward signalsfrom user feedback signals. Workbench 15 can implement a fine-tuning pipeline 17-3 to finetune development model 16.

[0134] 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.

[0135] 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.

[0136] In some implementations, pre-trained or fine-tuned models can achieve satisfactory performance without exemplars in the inputs. For instance, zero-shot prompts can include inputs that lack exemplars. Zero-shot prompts can be within a domain within a training dataset or outside of the training domain(s).

[0137] 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 on one or more training iterations. Workbench 15 can implement prompt engineering tools in development model 16.

[0138] 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.

[0139] 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.

[0140] 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 techniquesadapted for training a wide variety of machine-learned models. Example training techniques can correspond to the example training method 580 described above.

[0141] Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality' of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance quality where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution and pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models — e.g., understanding an intent in an unstructured request for a task — while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.

[0142] 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”).

[0143] 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.

[0144] 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 instructions that initiate API calls to send or obtain data via external svstems.

[0145] 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.

[0146] 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 and 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 16 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.

[0147] 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 16. Output model 20 can be a deployment version of development model 16. Output model 20 can be a development or training checkpoint of development model 16. Output model 20 can be a distilled, compressed, or otherwise optimized version of development model 16.

[0148] Figure 10 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 figures. 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. Figure 10 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 present disclosure. Figure 10 is described with reference to elements / terms described with respect to other systems and figures for exemplars' 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.

[0149] 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.

[0150] Initialized model 21 can undergo pre-training in a pre-training stage 22. Pretraining 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).

[0151] Pre-trained model 23 can then be anew 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 1 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 has satisfactory’ performance, if the model was already fine-tuned, or if other tuning approaches are preferred.

[0152] 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 dow nstream system(s) 28 for deployment or further development.

[0153] 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 can 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.

[0154] Figure 11 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.

[0155] 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.

[0156] 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 includepublic 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.

[0157] 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.

[0158] For example, model host 31 can operate on a server system that provides a machine-learning sendee 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.

[0159] In some implementations, model host 31 can operate on the 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 chent(s) 32. Model host 31 can be a part of the 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.

[0160] 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.

[0161] 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 dynamic 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 memory.

[0162] 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.

[0163] 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.

[0164] 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.

[0165] 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.

[0166] Model host 31 can access a library of pre-trained adapters or LoRA modules that can adapt a baseline model to align its outputs with a desired performance profile, augment model capabilities (e.g., to adapt to a different input modality, etc.), and the like. For instance, model host 31 can receive an input request to load a customized model, and model host 31 can retrieve one or more components to adapt a baseline model to the custom profile. Model host 31 can determine that a particular functionality is needed for a particular task (e.g., based on an output of a model that preprocesses an input) and retrieve a pre-trained component accordingly.

[0167] 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 and 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.

[0168] 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 moreimages, 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.

[0169] 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).

[0170] 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 anupscaled 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.

[0171] 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 can 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.

[0172] 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.

[0173] 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) I 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 visualizationoutput. 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.

[0174] 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. 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.

[0175] 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.

[0176] 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.

[0177] 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) 2can 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 process 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.

[0178] 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 modality7as 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.

[0179] 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).

[0180] 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 desired 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).

[0181] 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).

[0182] Figure 12 is a block diagram of an example networked computing system that can perform aspects of example implementations of the present 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 present 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 present 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 present 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, servercomputing system(s) 60, or model development platform system(s) 70 can interact in the performance of various aspects of the present disclosure (e.g., engaging third-party tools, accessing third-party databases or other resources, etc.).

[0183] 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 Figure 12 can be co-located with, contained by, or otherwise integrated into one or more other devices or systems.

[0184] 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).

[0185] 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.

[0186] 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 input object (e.g., a finger or a stylus). The touch-sensitive component can serve toimplement 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.

[0187] 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.

[0188] Server computing system(s) 60 can include one or more processors 61 and a memory762. 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.

[0189] 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.

[0190] Sen' er 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 can 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-learnedmodel(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.

[0191] 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 sendee, 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.

[0192] 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.

[0193] 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 includeone 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 wi th 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).

[0194] Figure 12illustrates one example arrangement of computing systems that can be used to implement the present 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).

[0195] Figure 13 is a block diagram of an example computing device 98 that performs according to example embodiments of the present 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 text messaging application, an email application, a dictation application, a virtual keyboard application, a browser application, etc. As illustrated in Figure 13, 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.

[0196] Figure 14 is a block diagram of an example computing device 99 that performs according to example embodiments of the present 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 text messaging application, an email application, a dictation application, a virtual keyboard application, a browser 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).

[0197] The central intelligence layer can include a number of machine-learned models. For example, as illustrated in Figure 14, 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.

[0198] 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 Figure 14, 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, the central device data layer can communicate with each device component using an API (e.g., a private API).

[0199] The computing systems illustrated in Figures 12 through 14 can be configured to perform the spatial computing and generative model interaction techniques described herein. These computing systems can include a wearable user device (e.g., smart glasses, a mixed-reality headset, or a virtual reality headset), a paired mobile device (e.g., a smartphone or tablet), or a backend server. The systems typically include a processor, memory, storage, and various interface bridges. These components can be interconnected to support data processing for multimodal interaction.

[0200] The processor can execute operations related to the described methods. For example, in the context of multi-modal input acquisition, the processor can execute onboardspeech-to-text engines to transcribe user voice inputs and detect trigger keywords (e.g., "this" or "that") in real-time. The processor can perform temporal correlation, aligning timestamps of detected keywords with gaze data to determine a point of attention. To support these operations, the processor can include or be coupled with hardware accelerators, such as Neural Processing Units (NPUs) or Tensor Processing Units (TPUs). These accelerators can operate to run local inference tasks, such as object detection or wake-word detection, before data is sent to a generative model.

[0201] The memory and storage devices can buffer and store data used for model input formation. When the system captures visual data corresponding to a gaze location, this image data or video buffering can be stored in the memory. The memory can also store data for constructing a structured data package (e.g., a JSON object) that combines transcription, visual data, and extracted text (e.g., via OCR). This arrangement facilitates the assembly of the multi-modal payload before transmission to a generative model system.

[0202] High-speed interfaces can manage data streams from system sensors. These interfaces can handle data from outward-facing cameras capturing a field of view and inwardfacing eye-tracking sensors. Low-speed interfaces can communicate with external peripherals or send control signals to electromechanical actuators. This can occur, for example, during an action execution phase (e.g., moving a physical object in response to a generated script).

[0203] In implementations where the computing system comprises a wearable device, input / output components can include a sensor array. This array can facilitate the monitoring steps of the described methods. The array can include eye-tracking sensors (e.g., infrared cameras and illuminators) that provide data for determining a gaze vector. The array can also include microphones for capturing voice commands. Additionally, Inertial Measurement Units (IMUs) can track head pose. This helps align the coordinate system of the eye-tracker with a real-world environment or a virtual overlay.

[0204] Communication interfaces can enable the system to function within a networked environment. These interfaces (e.g., wireless local area netw ork interfaces or cellular interfaces) allow the device to transmit a model input to a remotely hosted generative model system via API calls. Upon receiving a response, the processor can use the communication interface to receive resulting data or code. If the response involves executable computer code (e.g., scripting languages), the system can use stored firmware or software libraries to instantiate a runtime environment (e g., an interpreter). This environment can execute the generated code to perform a physical or digital action.

[0205] 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.

[0206] While the present subject matter has been described in detail with respect to various specific example embodiments thereof, 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 subject disclosure does not preclude inclusion of such modifications, variations or additions to the present subject matter as would be readily apparent to one of ordinary skill in the art. For instance, 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 present disclosure cover such alterations, variations, and equivalents.

[0207] 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 present 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 present subject matter as would be readily apparent to one of ordinary7skill 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 explanatory7purposes 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.”

[0208] 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 present disclosure.

[0209] 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 present disclosure.

Claims

WHAT IS CLAIMED IS:

1. A spatial computing method implemented by a computing system comprising a user device, wherein the method comprises:receiving user input comprising a user instruction, wherein the user input refers to one or more elements within a field of view of the user device;forming a model input for a generative model system based on the user input; sending the model input to the generative model system and obtaining a corresponding response; andperforming an action using the response.

2. The method of claim 1, comprising:monitoring a location of a point of attention of the user within the field of view, anddetermining one or more parts of the model input based on the location of the point of attention at one or more times.

3. The method of claim 2, wherein the location of the point of attention is determined using eye tracking.

4. The method of any one of the preceding claims, wherein the user device is a mixed reality device.

5. The method of any one of claims 1 to 3. wherein the user device is a virtual reality device.

6. The method of any one of claims 1 to 4, wherein the user device comprises a camera which defines the field of view.

7. The method of any one of the preceding claims, wherein the computing system comprises a computing device configured to perform the action, wherein the computing device is a different device to the user device.

8. The method of claim 7, wherein the computing device is within the field of view of the user device.

9. The method of claim 7 or claim 8, wherein the computing device comprises a display configured to display the result of the action and / or to display an indication that the action has been performed.

10. A method implemented by one or more computing devices, comprising:monitoring a location of a point of attention of a user within a field of view, wherein the location of the point of attention is monitored using eye-tracking;receiving user input comprising a user instruction, wherein the user input refers to one or more elements within the field of view;forming a model input for a generative model system based on the user input, comprising determining one or more parts of the model input based on the location of the point of attention at one or more times;sending the model input to the generative model system and obtaining a corresponding response, andperforming an action using the response.

11. The method of claim 10, wherein the location at the one or more times comprises a location associated with one of the elements.

12. The method of any one of the preceding claims, wherein the user input comprises a voice input.

13. The method of claim 12 when dependent on claim 2 or claim 10, wherein determining the one or more parts of the model input comprises temporally correlating the location of the point of attention with one or more trigger events.

14. The method of claim 13, comprising evaluating the voice input for the one or more trigger events.

15. The method of claim 14, wherein the one or more trigger events comprise the detection of one or more words in the voice input.

16. The method of any one of the preceding claims when dependent on claim 12. wherein the method comprises determining one or more parts of the model input based on the voice input.

17. The method of claim 16, wherein determining one or more parts of the model input based on the voice input comprises forming a transcription of at least part of the voice input and including the transcription in the model input.

18. The method of claim 16 or claim 17, wherein forming the model input comprises generating a plurality of tokens by tokenising the voice input, and including the plurality of tokens in the model input.

19. The method of any one of the preceding claims, wherein the model input comprises one or more timestamped parts.

20. The method of any one of the preceding claims, wherein the model input includes visual data corresponding to at least part of the field of view.

21. The method of claim 20. wherein the visual data comprises pixel data.

22. The method of claim 20 or claim 21, wherein the model input comprises image data for the entire field of view.

23. The method of any one of the preceding claims, wherein the user instruction comprises an instruction to generate an image based on one or more elements within the field of view, wherein the corresponding response comprises a generated image, and wherein performing an action using the response comprises displaying the generated image.

24. The method of any one of the preceding claims, wherein the method comprises obtaining computer code, wherein obtaining the computer code comprises sending the model input to the generative model and obtaining a corresponding response, and wherein performing the action using the response comprises executing the computer code.

25. The method of claim 24 wherein the user instruction comprises an instruction to perform an action in response to a condition being met.

26. The method of claim 25, wherein the condition comprises recognition of a particular gesture.

27. The method of claim 25 or claim 26 wherein the condition comprises recognition of a particular sound.

28. The method of any one of claims 24 to 27 wherein the user instruction comprises an instruction to control areal-world object.

29. The method of any one of claims 24 to 28, wherein the user instruction comprises an instruction to assign an action to areal-world object.

30. The method of any one of claims 24 to 29, wherein executing the computer code comprises carrying out the user instruction.

31. One or more non-transitory computer-readable media storing instructions that are executable by one or more processors to cause one or more computing devices to perform the method of any one of the preceding claims.

32. A computing system, comprising: one or more processors; and one or more non-transitory computer-readable media storing instructions that are executable by the one or more processors to cause one or more computing devices to perform operations, the operations comprising the method of any one of claims 1 to 30.

Images