Systems and methods for reducing image signal processor (ISP) latency in video acquisition and processing pipelines. In at least one embodiment, processing circuitry is provided in a video acquisition and processing pipeline to perform intelligent slicing of video frames for reducing ISP latency. In at least one embodiment, the intelligent slicing determines slice pixel height based on camera information and an ISP clock rate.
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
G06V 10/72 - Data preparation, e.g. statistical preprocessing of image or video features
G06V 10/94 - Hardware or software architectures specially adapted for image or video understanding
G06V 20/90 - Identifying an image sensor based on its output data
Approaches presented herein provide for automated management of operational systems, such as cooling and power systems, for a computing environment such as a data center. A secondary coolant distribution unit (CDU) can be connected to a cooling loop to allow for cutover from coolant flow from a primary CDU, which can allow for tasks such as coolant cleaning or replacement without having to power down the electronic components being cooled by the coolant flow. The cooling system can also be used to cure thermal interface materials (TIMs) by heating the flow of liquid through the cooling system to at least one target temperature over a period of time according to a curing profile. The electronic components can also have tags attached that allow for automatic detection and reading when installed in the electronic environment, which allows for automated detection and analysis of the needs of a set of components, as well as the automated adjustment of one or more operational systems to satisfy those needs with zero downtime of the impacted components.
Approaches presented herein provide for automated management of operational systems, such as cooling and power systems, for a computing environment such as a data center. A secondary coolant distribution unit (CDU) can be connected to a cooling loop to allow for cutover from coolant flow from a primary CDU, which can allow for tasks such as coolant cleaning or replacement without having to power down the electronic components being cooled by the coolant flow. The cooling system can also be used to cure thermal interface materials (TIMs) by heating the flow of liquid through the cooling system to at least one target temperature over a period of time according to a curing profile. The electronic components can also have tags attached that allow for automatic detection and reading when installed in the electronic environment, which allows for automated detection and analysis of the needs of a set of components, as well as the automated adjustment of one or more operational systems to satisfy those needs with zero downtime of the impacted components.
In one embodiment, a target system, includes a network interface to receive packets from at least one initiator, packet processing circuitry to process the packets according to work requests stored in a receive queue in memory, generate completion notifications responsively to completing processing of corresponding ones of the work requests, and spray the completion notifications of a given queue pair among different completion queues for processing by corresponding processing cores.
In various examples, techniques for virtual sensor outputs for neural rendering systems and applications is described herein. For instance, systems and methods described herein may generate virtual data-such as virtual sensor data representing virtual outputs corresponding to depth sensors-to train one or more neural networks associated with performing neural rendering. For example, sensor data representing an environment may be obtained, such as image data, LiDAR data, and/or any other type of sensor data. Virtual data may then be generated for a portion of the sensor data that is associated with a surface and/or an object, such as a driving surface for which limited information is usually provided by the sensor data. The neural network(s) associated with performing neural rendering may then be trained using both the sensor data along with the virtual data.
G06T 7/80 - Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Disclosed are apparatuses, systems, and techniques for implementing efficient streaming diarization of multi-speaker speech. The techniques include processing, using a trained speaker diarization model, an input that includes (i) one or more digital units (DUs) of a segment of speech and (ii) a speaker cache of one or more previously processed DUs of the speech to obtain a plurality of likelihood values. Each likelihood value characterizes a likelihood that one or more speakers are speaking in the segment of the speech. The speaker cache is mapped to an order of appearance, in the speech, of one or more previously detected speakers. The techniques further include updating the speaker cache using at least the one or more DUs of the segment of the speech and the plurality of likelihood values. The techniques further include generating, using the plurality of likelihood values, a speaker-to-time mapping for the speech.
G10L 21/0308 - Voice signal separating characterised by the type of parameter measurement, e.g. correlation techniques, zero crossing techniques or predictive techniques
G10L 17/04 - Training, enrolment or model building
G10L 25/18 - Speech or voice analysis techniques not restricted to a single one of groups characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
G10L 25/60 - Speech or voice analysis techniques not restricted to a single one of groups specially adapted for particular use for comparison or discrimination for measuring the quality of voice signals
G10L 25/78 - Detection of presence or absence of voice signals
7.
TELEPORTATION SYSTEM COMBINING VIRTUAL REALITY AND AUGMENTED REALITY
Apparatuses, systems, and techniques providing a teleportation system combining virtual reality and augment reality are provided. A first set of data associated with a real-world environment is received. A virtual object candidate corresponding to an object in the real-world environment positioned at a first location is identified based on a subset of the first set of data. A second location of a first user within the real-world environment is identified based on the first set of data. A second set of data representing a first avatar of a guest user is received. A virtual representation of the real-world environment is generated based on the first and second sets of data, comprising the virtual object candidate positioned at the first location and a second avatar of the first user and/or the first avatar of the guest user. The virtual representation is send to a computing system associated with the guest user.
Apparatuses, systems, and techniques to generate computer graphics. In at least one embodiment, an application programming interface call to output an application-generated frame of computer graphics is intercepted. One or more interpolated frames of computer graphics are generated based on the application-generated frames. The application-generated and interpolated frames are output in accordance with a goal rate.
In various examples, a generalizable mobility model can receive a state and identifier of a robot, and generate an action for the robot based on the state and the identifier. The state can identify a position, environment, and navigation goal of the robot while the identifier can indicate a type of the robot. The generalizable mobility model can use the identifier to generate an action for the robot to reach the navigation goal from its position based on the type of the robot, such as a humanoid, quadruped, or wheeled robot. The generalizable mobility model can be a distilled combination of multiple robot type-specific models, and can use the identifier to mimic type-specific actions output by the multiple robot type-specific models to transmit to the robot to move the robot.
In various examples, a system can include one or more processors to segment, using at least one machine learning model, a first representation of an object into a plurality of parts according to a hierarchy, where the hierarchy indicates groups of datapoints for at least one part of the plurality of parts based on a value. The one or more processors generate at least one second representation of the object based on the at least one of the part of the plurality of parts.
G06V 10/762 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
One or more embodiments of the present disclosure relate to executing, by a plurality of compute engines, a plurality of runnables of a computing application based at least on an execution schedule and a set of commands associated with the execution schedule. The execution schedule may be generated using a compiling system to include the set of commands. The set of commands may include one or more individual commands corresponding to one or more timing fences dictating a timing and order of execution of one or more individual runnables of the plurality of runnables.
G06F 11/07 - Responding to the occurrence of a fault, e.g. fault tolerance
G06F 21/52 - Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems during program execution, e.g. stack integrity, buffer overflow or preventing unwanted data erasure
12.
MESH TOPOLOGY GENERATION USING PARALLEL PROCESSING
Various embodiments include techniques for generating topological data for a mesh included in a computer-generated environment. The mesh includes simple geometric shapes, such as triangles. The disclosed techniques identify vertices in the mesh that have the same position and have identical attributes, such as color, normal vector, and texture coordinates. The disclosed techniques further identify vertices in the mesh that have the same position but differ in one or more attributes. The techniques generate lists of the triangles that are adjacent to each vertex included in the mesh. The techniques generate a list of the unique edges included in the mesh. Further, the techniques are well suited for execution on highly parallel processors, such as graphics processing units, thereby reducing the time to generate this topological data. The topological data may then be efficiently used by other computer graphics processing operations.
Approaches presented herein provide for the reconstruction of implicit multi-dimensional shapes. In one embodiment, oriented point cloud data representative of an object can be obtained using a physical scanning process. The point cloud data can be provided as input to a trained density model that can infer density functions for various points. The points can be mapped to a voxel hierarchy, allowing density functions to be determined for those voxels at the various levels that are associated with at least one point of the input point cloud. Contribution weights can be determined for the various density functions for the sparse voxel hierarchy, and the weighted density functions combined to obtain a density field. The density field can be evaluated to generate a geometric mesh where points having a zero, or near-zero, value are determined to contribute to the surface of the object.
Disclosed are apparatuses, systems, and techniques for implementing efficient transcription of multi-speaker speech with overlapping utterances using speaker activity detection. The techniques include processing, using a first set of neural network (NN) layers of an automatic speech recognition (ASR) model, speech data for the multi-speaker speech to generate an intermediate feature (IF) representative of the speech data. The techniques further include modifying, using a second set of NN layers of the ASR model, the IF to obtain modified IFs using speaker activity data, which identifies times when various speakers speak in the multi-speaker speech. The techniques further include processing the modified IFs to obtain a plurality of transcriptions identifying content of speech of the plurality of speakers, and generating, using the plurality of transcriptions, a transcript of the multi-speaker speech.
In various examples, a single generalizable mobility model can be applied to a variety of robot types. Actions output by the generalizable mobility model can be based on actions generated using imitation learning, and the actions can be refined per type of robot using residual reinforcement learning. A type-specific model can be updated per type of robot using residual reinforcement learning, and can be combined and distilled into the single generalizable model. Actions output by the type-specific model can be combined with actions generated using imitation learning, and results of the action on a simulated robot can update weights of one layer of the type-specific model. By building upon an imitation learning model and refining each type-specific model using residual reinforcement learning, the generalizable mobility model can improve data and time efficiency compared to continuous multi-type training while also avoiding joint multi-type training complexities.
Disclosed are apparatuses, systems, and techniques for trusted training and deployment of artificial intelligence (AI) for generating media content that uses likeness to natural persons and other protected objects. The techniques include receiving, from a client device by a cloud server, a prompt for a media content associated with likeness to a protected object and authenticating access, associated with the prompt, to likeness data for the protected object. The techniques further include loading, into a secure execution container, generative media content model(s) and a representation of the likeness data and processing, in the secure execution container and using the representation of the likeness data and the generative media content model(s), the prompt to generate the media content. The techniques further include causing, responsive to instructions associated with the prompt, communicating the media content and/or storing the media content.
In various examples, at least partial control of a vehicle may be transferred to a control system remote from the vehicle. Sensor data may be received from a sensor(s) of the vehicle and the sensor data may be encoded to generate encoded sensor data. The encoded sensor data may be transmitted to the control system for display on a virtual reality headset of the control system. Control data may be received by the vehicle and from the control system that may be representative of a control input(s) from the control system, and actuation by an actuation component(s) of the vehicle may be caused based on the control input.
Approaches presented herein provide for automated management of operational systems, such as cooling and power systems, for a computing environment such as a data center. A secondary coolant distribution unit (CDU) can be connected to a cooling loop to allow for cutover from coolant flow from a primary CDU, which can allow for tasks such as coolant cleaning or replacement without having to power down the electronic components being cooled by the coolant flow. The cooling system can also be used to cure thermal interface materials (TIMs) by heating the flow of liquid through the cooling system to at least one target temperature over a period of time according to a curing profile. The electronic components can also have tags attached that allow for automatic detection and reading when installed in the electronic environment, which allows for automated detection and analysis of the needs of a set of components, as well as the automated adjustment of one or more operational systems to satisfy those needs with zero downtime of the impacted components.
H05K 7/20 - Modifications to facilitate cooling, ventilating, or heating
G06K 7/10 - Methods or arrangements for sensing record carriers by electromagnetic radiation, e.g. optical sensingMethods or arrangements for sensing record carriers by corpuscular radiation
Techniques for thermal interface material testing are described. In one example, a thermal interface material testing device includes a bimaterial strip, a pressure plate that applies a pressure to a thermal interface material between the pressure plate and the bimaterial strip, one or more thermal cycling devices that perform thermal cycling of the bimaterial strip to test the thermal interface material, and one or more sensor devices that generate sensor data based on the thermal cycling of the bimaterial strip.
Physically unclonable function cells based on bit-storing machine-memory circuits include a pair of parallel circuit branches, each of the parallel branches including a first transistor configured to be always-ON when the physically unclonable function cell is powered, and a second transistor cross-coupled to an opposite one of the parallel branches.
H03K 19/17768 - Structural details of configuration resources for security
H03K 19/0944 - Logic circuits, i.e. having at least two inputs acting on one outputInverting circuits using specified components using semiconductor devices using field-effect transistors using MOSFET
H03K 19/17784 - Structural details for adapting physical parameters for supply voltage
21.
CLOUD-HOSTED MANAGEMENT FOR EDGE COMPUTING DEVICES
Systems and methods include a cloud-hosted management plane where an authenticated user can initiate a request to establish a remote session to an edge device. A request may be transmitted to create a secure tunnel between the edge device and the cloud-based management plane. The edge device may receive and use a set of credentials to trigger a web-socket tunnel from the edge device to the cloud-hosted management plane. This connection may be used for management of an edge device(s), troubleshooting edge inferencing, access/use of edge applications, and various other applications.
An optical transmitter includes lasers configured to generate a wave division multiplex (WDM) on a light guide, and a Code Division Multiple Access (CDMA) symbol generator coupled to modulate CDMA symbols on the light guide across the channels of the WDM. The transmitter utilizes of laser locking controls configured to correlate the CDMA symbols to frequency adjustments applied to the lasers.
Apparatuses, systems, and techniques to allocate power to one or more processors. In at least one embodiment, processors or computing systems perform an API to allocate power to one or more processors based, at least in part, on indications of priority of one or more threads to be performed by the one or more processors.
Embodiments relate to hazard detection in autonomous and semi-autonomous systems and applications. A transformer may use sampled image and LiDAR features to extract and decode a representation of one or more features of each point (e.g., refined height, range, driving condition, etc.) on a sampled surface (e.g., the road). These detections may be provided to one or more control components of an autonomous vehicle, which may use the detections to navigate, plan, or otherwise perform one or more operations. Some embodiments employ an automated approach to derive ground truth data from sensor data collected by data collection vehicle(s), such as data representing detected ground surface models, detected surface features, detected weather and/or surface condition labels, and/or detected per-point artifact labels. Accordingly, surface features such as ground surface heights along a predicted trajectory may be detected and ground truth data may be generated for a variety of sensing tasks.
B60W 60/00 - Drive control systems specially adapted for autonomous road vehicles
G01S 17/89 - Lidar systems, specially adapted for specific applications for mapping or imaging
G06V 10/80 - Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
G06V 10/82 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
25.
SYNTHETIC EYE IMAGE GENERATION USING NEURAL NETWORKS
One embodiment of a method includes calculating one or more activation values of one or more neural networks trained to infer eye gaze information based, at least in part, on eye position of one or more images of one or more faces indicated by an infrared light reflection from the one or more images.
G06V 10/774 - Generating sets of training patternsBootstrap methods, e.g. bagging or boosting
G06F 3/01 - Input arrangements or combined input and output arrangements for interaction between user and computer
G06F 7/57 - Arithmetic logic units [ALU], i.e. arrangements or devices for performing two or more of the operations covered by groups or for performing logical operations
A vertical patterned magnetic flux inductor system includes a magnetic core volume that defines a longitudinal gap extending horizontally through the center of the core volume. Top and bottom members of the core volume have slots therein. The slots are arranged such that the top and bottom members define complementary S shaped profiles when viewed in the z direction. Longitudinal electrical conductors are disposed partially in the slots and slope from the bottom member of the core to the top member of the core. A first terminal end of each conductor is exposed at the top side of the core volume and a second terminal end of each conductor is exposed at the bottom side of the core volume.
A horizontal patterned magnetic flux inductor system includes a magnetic inductor core that defines a longitudinal gap passing through the core volume in a vertical direction. The inductor core forms a three-dimensional flux path around the longitudinal gap whose projection onto the x-z plane forms a figure eight pattern with its two lobes oriented in the x direction. First and second electrical conductors are spaced apart in the x direction and pass through the core volume and through the longitudinal gap such that the inductor core causes inverse coupling between the conductors when electric current flows through the conductors in the same direction but causes positive coupling between the conductors when electric current flows through the conductors in opposite directions. Each of the electrical conductors has a front end exposed at the front side of the core volume and a back end exposed at the back side of the core volume.
H02M 1/32 - Means for protecting converters other than by automatic disconnection
H02M 3/158 - Conversion of DC power input into DC power output without intermediate conversion into AC by static converters using discharge tubes with control electrode or semiconductor devices with control electrode using devices of a triode or transistor type requiring continuous application of a control signal using semiconductor devices only with automatic control of output voltage or current, e.g. switching regulators including plural semiconductor devices as final control devices for a single load
An optical communication system includes at least one wave division multiplex (WDM) transmitter configured to generate WDM signals in multiple channels on a light guide, and a Code Division Multiple Access (CDMA) symbol generator coupled to modulate output of the WDM transmitter at a frequency below a noise floor of the WDM signals.
Disclosed are systems and techniques for efficient release of computing resources across memory contexts and processor pipelines. The techniques include acquiring a first lock and providing first work to be executed by a first processor. The techniques further include receiving a first signal from the first processor that the first lock can be released. The techniques further include providing the first signal to a second processor and releasing, by the second processor, the first lock.
In various examples, a technique for efficiently verifying sensor synchronization includes generating an index for a data sample from a sensor based on a timestamp associated with the data sample and a temporal resolution for buckets of data associated with different indices; retrieving a plurality of data samples from one or more other sensors based on a bucket associated with the index and one or more adjacent buckets; determining that a number of samples in the retrieved plurality of data samples is less than a number of sensors from which data samples are obtained; and based on determining that the number of samples in the retrieved plurality of data samples is less than the number of sensors from which data samples are obtained, performing one or more actions with respect to the data sample.
In various examples, systems and methods are disclosed relating to implementing language models for spatial sound synthesis. A system can identify a scene comprising at least one object or environmental feature. The system can determine, using a language model, a selection of an audio sample for the at least one object or environmental feature. The system can generate an updated scene including the audio sample associated with the at least one object or environmental feature.
Various embodiments include techniques for reducing IR drop in an electronic circuit. IR drop results from simultaneous transitioning of a large number of electronic signals of an interconnect in a densely packed area. This IR drop can result in data errors and malfunctioning circuit components. To mitigate such IR drops, the disclosed techniques stagger the electronic signals such that one half of the signals transition at the positive edge of a synchronizing clock signal and one half of the signals transition at the negative edge of the synchronizing clock signal. Further, selective data bus inversion is applied to each half of the electronic signals. When applied together, these two techniques limit simultaneous transition to 25% of the total number of the electronic signals of the interconnect, thereby reducing IR drop relative to prior conventional approaches.
In various examples, live perception from sensors of a vehicle may be leveraged to detect and classify intersection contention areas in an environment of a vehicle in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute outputs—such as signed distance functions—that may correspond to locations of boundaries delineating intersection contention areas. The signed distance functions may be decoded and/or post-processed to determine instance segmentation masks representing locations and classifications of intersection areas or regions. The locations of the intersections areas or regions may be generated in image-space and converted to world-space coordinates to aid an autonomous or semi-autonomous vehicle in navigating intersections according to rules of the road, traffic priority considerations, and/or the like.
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
G06V 10/82 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
G06V 20/56 - Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
Apparatuses, system, and techniques to process resources used to perform a neural network to generate one or more images. In at least one embodiment, a processor comprising circuitry uses one or more neural networks to generate one or more second images based, at least in part, on noise within a first image.
A neural network is trained to recognize an object based on a first portion trained adversarially and a second portion trained non-adversarially. During adversarial training, the first portion encodes input to a set of latent variables, and the second portion reconstructs the input by decoding the variables. The first portion is trained based on an error signal generated from the decoded output of the second portion, whose configuration is kept fixed during the adversarial training.
G06V 20/58 - Recognition of moving objects or obstacles, e.g. vehicles or pedestriansRecognition of traffic objects, e.g. traffic signs, traffic lights or roads
Apparatuses, systems, and techniques to identify objects in an image. In at least one embodiment, one or more neural networks identify the objects using pose information of the objects and visible properties of the objects.
Apparatuses, systems, and techniques to animate objects in computer-generated graphics. In at least one embodiment, one or more neural networks are trained to identify one or more forces to be applied to one or more objects based, at least in part, on training data corresponding to two or more aspects of motion of the one or more objects.
Apparatuses, systems, and techniques to identify users or sessions impacted by one or more issues of one or more software programs, based, at least in part, on characteristics shared by the users and other users of the programs who have indicated the one or more issues.
Disclosed are apparatuses, systems, and techniques that implement software-agnostic diagnostic of nodes within a deep learning environment. Monitoring agents within a node and external to a node can track metrics according to a policy provided by a controller. Upon identification of a metric violating a policy criterion, the monitoring agents can provide a policy violation notification to the controller. The controller can use the policy violation notification to determine the reliability of the node. Upon determination that the node is unreliable, the node may be prevented from receiving future task processing assignments.
In various examples, using artificial intelligence to alter content for streaming systems and applications is described herein. Systems and methods are disclosed that allow for users to update visual characteristics of applications—such as the contrast, saturation, brightness, luminance, color, texture, and/or the like—in real-time using one or more machine learning models. For instance, a user may provide input associated with updating a visual characteristic, such as in the form of speech and/or text. The machine learning model(s) may then process the input along with an image of the application to generate and provide sample images associated with the updated visual characteristic. Additionally, based on a selection of sample image, the machine learning model(s) may use at least the selected image to generate visual data (e.g., shader code) representing parameters for updating the visual characteristic for additional images rendered for the application.
G06V 10/75 - Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video featuresCoarse-fine approaches, e.g. multi-scale approachesImage or video pattern matchingProximity measures in feature spaces using context analysisSelection of dictionaries
G06V 10/54 - Extraction of image or video features relating to texture
G06V 10/56 - Extraction of image or video features relating to colour
G06V 10/60 - Extraction of image or video features relating to illumination properties, e.g. using a reflectance or lighting model
Approaches presented herein provide for receiving liquid coolant into a server or other liquid-cooled computer system at a cold plate having internal channels specifically routed to provide cooling to one or more identified component locations. The cold plate can be the primary source of liquid-cooling for the server, allowing for the removal of extraneous connections and hoses. In one embodiment, a cold plate can be fluidly sealed between an inlet and outlet of coolant into a server to prevent leaks onto damageable components. Compressible bellows extending out from an external surface of a cold plate along the channels can thermally interface with components at different heights. Such a cold plate can include flow control features within the channels, such as to fluidly seal the bellow or to provide microchannels at component interface locations.
In various examples, a technique for executing a systems engineering workflow includes matching a user input to graph data associated with an engineered system and determining a context based at least on one or more content items associated with the graph data. The technique also includes generating, via execution of a first machine learning model, an initial version of an additional content item based at least on the context and generating, via execution of a second machine learning model, one or more revisions to the additional content item based at least on one or more critiques associated with the additional content item. The technique further includes causing the engineered system to be updated based at least on the revision(s) to the additional content item.
In various examples, systems and methods are provided for obtaining predictions from one or more neural networks. For example, one or more draft sets of one or more tokens may be selected for input into one or more neural networks based at least on multiple amounts of processing time and multiple numbers of accepted tokens associated with multiple numbers of tokens. The one or more neural networks may then use the one or more draft sets to generate one or more predictions.
Systems and methods are disclosed related to synchronizing a monotonic signal. A monotonic signal indicates an event at which the signal changes in only one direction (either rising or falling edges). The signal is restored before each event and the falling or rising edge to restore the signal may occur asynchronously. In at least one embodiment, an asynchronous signal generated in a first clock domain is sampled in a second clock domain by a plurality of flip-flops in parallel and the sampled signals output by the plurality of flip-flops are combined produce a synchronized signal in the second clock domain. The number of sampling flip-flops may be increased to reduce the probability of the combined output remaining metastable to an acceptable level. A monotonic synchronizer circuit may be used to construct a first-in first-out (FIFO) synchronizer.
H03K 5/135 - Arrangements having a single output and transforming input signals into pulses delivered at desired time intervals by the use of time reference signals, e.g. clock signals
An autoregressive (AR) tokenizer can be a transformer-based architecture that tokenizes high-resolution, long-duration videos with a compression ratio and preserved reconstruction quality. Instead of directly quantizing inputs, the disclosed processes can use continuous tokens from the previous frame as inputs for the current frame. By bypassing the quantization layer for these continuous tokens, the video quality can be enhanced during reconstruction compared to conventional solutions. AR tokenizer can exploit the temporal causality between video frames by conditioning each frame's decoding on previous frames'pre-bottleneck features, reducing the temporal redundancy, and improving the reconstruction quality. While the performance of conventional transformer-based discrete tokenizers plateaus with increased training compute, the disclosed AR tokenizer can continue to improve, achieving an increase of 8 times the compression rate compared to conventional reconstruction algorithms.
An apparatus includes at least one heat pipe that is adapted to be thermally coupled to an integrated circuit and has an evaporator portion and a first condenser portion, wherein the first condenser portion extends away from the evaporator portion; a first plurality of cooling fins that is attached to the first condenser portion; a first movable support that is thermally coupled to the first condenser portion and is configured to move a second plurality of cooling fins relative to the first plurality of cooling fins; and the second plurality of cooling fins, which is attached to the first movable support.
In various examples, temporal-based perception for autonomous or semi-autonomous systems and applications is described. Systems and methods are disclosed that use a machine learning model (MLM) to intrinsically fuse feature maps associated with different sensors and different instances in time. To generate a feature map, image data generated using image sensors (e.g., cameras) located around a vehicle are processed using a MLM that is trained to generate the feature map. The MLM may then fuse the feature maps in order to generate a final feature map associated with a current instance in time. The feature maps associated with the previous instances in time may be preprocessed using one or more layers of the MLM, where the one or more layers are associated with performing temporal transformation before the fusion is performed. The MLM may then use the final feature map to generate one or more outputs.
B60W 60/00 - Drive control systems specially adapted for autonomous road vehicles
G06V 20/58 - Recognition of moving objects or obstacles, e.g. vehicles or pedestriansRecognition of traffic objects, e.g. traffic signs, traffic lights or roads
48.
DENOISING GEOMETRICALLY REFINED PARAMETRIC SURFACES USING MOTION VECTORS
In various examples, images of geometrically refined parametric surfaces may be denoised using motion vectors computed in 2D screen space and/or using locations of corresponding control points. Surfaces in a scene may be refined using any desired technique (e.g., subdivision or tessellation), an image of the scene may be ray traced or rasterized, and motion vectors may be computed in 2D screen space and/or using corresponding control points on the control cages for geometrically refined parametric surfaces. For example, the motion of a rendered (e.g., limit, displaced) surface may be computed by reconstructing current and previous 3D locations on the limit surface using the parametric location of the limit surface each pixel views, and the current and previous 3D locations may be projected into screen space and compared to generate a corresponding motion vector. As such, the image may be denoised using the motion vectors.
Mechanisms for implementing text-to-volumetric video transformers utilizing a first stage configured to transform a text prompt into a three-dimensional neural radiance field, and a second stage configured to apply a four-dimensional multi-resolution deformation field to generate time-varying displacements for the three-dimensional neural radiance field consistent with motion expressed in the text prompt, while maintaining a content of the three-dimensional neural radiance field static.
In various examples, a technique for evaluating a foundation model includes evaluating one or more versions of a robotics system foundation model based on one or more specified metrics and model data including simulated and/or real-world robotics system data. Each of the one or more versions of the robotics system foundation model may be represented by a checkpoint that describes the state of the foundation model after a given period of training, and/or after training using a specific set of hyperparameters. The technique may identify one or more winning checkpoints from a set of candidate checkpoints included in the model data. The technique may then evaluate the winning checkpoints based on specified metrics and test data included in the model data. The disclosed techniques may display the evaluation results to a user via one or more visualization tools, and/or store the evaluation results for later retrieval and/or processing.
Systems and methods disclosed herein may be used to provide rigid connectors using a circuit board assembly having a printed circuit board (PCB) with surface contact pads. The surface contact pads may be individually encompassed by rigid connector forms having individual external flanges. A rigid frame may overlay the external flanges to maintain a position for the rigid connector forms with respect to the PCB.
H01R 12/79 - Coupling devices for flexible printed circuits, flat or ribbon cables or like structures connecting to rigid printed circuits or like structures
H05K 7/14 - Mounting supporting structure in casing or on frame or rack
52.
MACHINE LEARNING MODEL WRAPPER FOR PREPROCESSING COMPLEX INSTRUCTIONS
In machine learning, some models, such as language models, are trained to perform language processing tasks. These language processing tasks can include text generation, machine translation, sentiment analysis, question answering, content generation, code development, etc. However, these models are inherently unreliable, especially when prompted with complex (e.g. composite) instructions. For example, they are prone to miscarrying basic operations, ignoring explicitly stated formatting and behavioral constraints, and hallucinating new steps or step outcomes. The present disclosure provides a mechanism that preprocesses an instruction to a machine learning model to generate a sequence of instructions via which flow control and data control in the model may be enforced, thereby enhancing reliability of the output generated by the model.
In various examples, machine learning-based sign language translation for video conferencing platforms is provided. A video conferencing plug-in for sign language communication may instantiate sign language translation modules that translate incoming communication channel data from a first language mode (e.g., spoken language or sign language) to a target language mode corresponding to a sign language preference for the user. The video conferencing plug-in for sign language communication may control a user interface for a client application for the video conferencing platform to present at least a portion of an avatar performing signing corresponding to the sign language translation data based on control data generated by the video conferencing plug-in for sign language communication.
Apparatuses, systems, and techniques to determine orientation of an objects in an image. In at least one embodiment, images are processed using a neural network trained to determine orientation of an object.
G06V 10/24 - Aligning, centring, orientation detection or correction of the image
G05D 1/00 - Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
G05D 1/249 - Arrangements for determining position or orientation using signals provided by artificial sources external to the vehicle, e.g. navigation beacons from positioning sensors located off-board the vehicle, e.g. from cameras
G05D 101/20 - Details of software or hardware architectures used for the control of position using external object recognition
Apparatuses, systems, and techniques to select a neural network architecture from a plurality of neural networks in a federated learning (FL) setting. In at least one embodiment, a neural network is trained by combining training results from different FL computing systems, where each of the different FL computing systems, for example, trains different portions of the neural network.
Apparatuses, systems, and techniques to perform an application programming interface (API) to add one or more graph nodes to a software graph, wherein the API is to store an indication of whether a node within a software graph was performed based, at least in part, on a dependency type indicated by the API. In at least one embodiment, one or more nodes are added to a graph in accordance to one or more dependency types.
Apparatuses, systems, and techniques to execute programs in a single hardware context on a graphics processing unit (GPU). In at least one embodiment, resource management patches expressed in library or executable code are applied to one or more kernels to ensure execution in a shared context on a GPU.
Apparatuses, systems, and techniques are presented to generate image data. In at least one embodiment, one or more neural networks are used to cause a lighting effect to be applied to one or more objects within one or more images based, at least in part, on synthetically generated images of the one or more objects.
Class agnostic object mask generation uses a vision transformer-based auto-labeling framework requiring only images and object bounding boxes to generate object (segmentation) masks. The generated object masks, images, and object labels may then be used to train instance segmentation models or other neural networks to localize and segment objects with pixel-level accuracy.
In various examples, at least partial control of a vehicle may be transferred to a control system remote from the vehicle. Sensor data may be received from a sensor(s) of the vehicle and the sensor data may be encoded to generate encoded sensor data. The encoded sensor data may be transmitted to the control system for display on a virtual reality headset of the control system. Control data may be received by the vehicle and from the control system that may be representative of a control input(s) from the control system, and actuation by an actuation component(s) of the vehicle may be caused based on the control input.
In various examples, machine learning data mining for autonomous or semi-autonomous systems and applications is described herein. Systems and methods are disclosed that use neural networks to perform one or more data mining processes. For instance, a first neural network(s) may process input data (e.g., image data) to remove data samples (e.g., images) that are associated with a first object classification(s) and/or a second neural network(s) may process the input data to retrieve data samples (e.g., images) that are associated with a second classification(s). Next, a third neural network(s) may process filtered input data (e.g., the input data not removed by the first neural network(s) and/or the input data retrieved by the second neural network(s)) to determine uncertainty classifications associated with the data samples and a fourth neural network(s) may process the filtered input data to determine final object classifications associated with the data samples.
Apparatuses, systems, and techniques to transpose a bit matrix by performing threads concurrently. In at least one embodiment, a processor comprising one or more circuits causes one or more matrix operands to be transposed by loading two or more portions of those one or more matrix operands by two or more software threads concurrently.
Apparatuses, systems, and techniques to identify information to evict from a Key-Value (KV) cache. In at least one embodiment, information stored within one or more large language model (LLM) KV caches may be identified to cause an indication to be generated of information stored within the one or more LLM KV caches that can be evicted without causing other information stored within the one or more LLM KV caches to be restored to the one or more LLM KV caches.
G06F 12/0875 - Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches with dedicated cache, e.g. instruction or stack
G06F 12/126 - Replacement control using replacement algorithms with special data handling, e.g. priority of data or instructions, handling errors or pinning
09 - Scientific and electric apparatus and instruments
42 - Scientific, technological and industrial services, research and design
Goods & Services
Computer hardware, namely, computer storage systems for extending GPU memory with network-attached flash storage optimized for artificial intelligence inference workloads; computer hardware, namely, computer hardware for accelerating data storage and retrieval in artificial intelligence inference systems; computer hardware, namely, computer hardware featuring integrated processors for software-defined acceleration of storage, networking, and security functions; computer hardware, namely, hardware accelerators for facilitating distributed key-value cache management in artificial intelligence computing systems; computer hardware for enabling implementation of reference architectures for artificial intelligence (AI)-native storage infrastructure, context memory storage, and distributed key-value cache management for AI inference systems; computer servers, namely, storage servers and storage enclosures for managing context memory in artificial intelligence computing pods; data storage devices, namely, computer storage platforms comprising computer hardware and software sold as a unit for storing, managing, and routing key-value cache data for artificial intelligence inference computing systems; data storage devices, namely, data storage apparatus for storing and managing inference context data in artificial intelligence computing systems; downloadable software for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; downloadable software, namely, software libraries, application programming interfaces (APIs), and software development kits (SDKs) for implementing reference architectures for artificial intelligence (AI)-native storage infrastructure and context memory storage systems; downloadable software for enabling implementation of reference architectures for artificial intelligence (AI)-native storage infrastructure, context memory storage, and distributed key-value cache management for AI inference systems; computer software, namely, device drivers for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; downloadable software development kits (SDKs) for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; downloadable electronic publications in the nature of technical specifications, reference architecture documents, implementation guides, and technical manuals in the field of artificial intelligence (AI)-native storage infrastructure, context memory storage, and management of key-value cache data and inference context for AI inference systems; downloadable application programming interfaces (APIs) for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; downloadable computer software libraries for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; computer hardware for accelerating artificial intelligence (AI) inference workflows; computer hardware featuring integrated processors for software-defined acceleration of storage, networking, and security functions; hardware accelerators for facilitating distributed AI-native key-value (KV) cache management; hardware and downloadable software sold as a unit for managing agentic workflows and treating inference context as a persistent data type Providing temporary use of online non-downloadable software for managing, routing, storing, retrieving, sharing, and transferring key-value cache data and inference context across artificial intelligence compute, networking, and storage systems; providing online non-downloadable software for the storage of electronic data; providing online non-downloadable software for managing large scale data storage and storage system performance analysis; electronic storage of data; computer software technical support services, technical information and technical support regarding software patches, upgrades and updates; providing information relating to computer technology; providing information relating to computer technology, namely, providing technical information, consulting, design, and development services relating to reference architectures for artificial intelligence (AI)-native storage infrastructure, context memory storage, and distributed key-value cache management for AI inference systems; providing information relating to computer technology, namely, providing support services for implementing artificial intelligence (AI)-native storage reference architectures; providing information relating to computer technology, namely, troubleshooting and optimization of software-defined acceleration for storage, networking, security, and inference context management; Technology consultation in the field of artificial intelligence (AI); providing computer hardware and software information online; providing information in the fields of technology and software development via an on-line website; providing information relating to computer technology and programming via a website; software as a service (SaaS) featuring software for managing agentic workflows in artificial intelligence (AI) applications; providing temporary use of non-downloadable software for managing, distributing, and optimizing access to AI-native key-value (KV) cache data within agent-based inference systems; technical support services in the field of AI-native storage infrastructure, namely, troubleshooting performance of software-defined acceleration for storage, networking, and security; design and development of custom hardware and software solutions for managing persistent inference context in AI-native data workflows; consulting services related to the integration of agentic workflow orchestration and AI-native storage systems
65.
Application programming interface to indicate kernel dependencies
Apparatuses, systems, and techniques to perform an application programming interface (API) to add one or more graph nodes to a software graph, wherein the API is to cause a kernel node to be added to a software graph based, at least in part, on a dependency type indicated by the API. In at least one embodiment, one or more nodes are added to a graph in accordance to one or more dependency types.
Embodiments described herein provide a hollow L-shaped structure used as a busbar for distributing electrical power in a data center. In at least one embodiment, the hollow L-shaped busbar may be made by folding a trapezoidal sheet of conductive metal.
Apparatuses, systems, and techniques to provide emotion-based experience adjustment. In at least one embodiment, one or more neural networks are used to adjust one or more properties of one or more games based on an emotional state of a user.
A63F 13/67 - Generating or modifying game content before or while executing the game program, e.g. authoring tools specially adapted for game development or game-integrated level editor adaptively or by learning from player actions, e.g. skill level adjustment or by storing successful combat sequences for re-use
A63F 13/79 - Game security or game management aspects involving player-related data, e.g. identities, accounts, preferences or play histories
Apparatuses, systems, and techniques to generate a three-dimensional (3D) information. In at least one embodiment, the 3D information is generated using one or more neural networks and comprises, for example, one or more map features.
One embodiment of a method for padding data includes storing one or more memory addresses at which padding is to be added in a queue; determining whether a first memory address is stored in the queue; if the first memory address is stored in the queue, then writing a zero to a register; if the first memory address is not stored in the queue, then writing a first value that is read from a first memory to the register; and performing at least one computation based on the first value that is written to the register to generate an output.
In various examples, machine learning model-based sign language corrective feedback is provided. Machine learning model-based recognition of sign language symbols (e.g., body poses and/or movements) may be used to generate real-time kinematic feedback to a sign language speaker that guides them to adjust their hand pose to correctly align with established sign language standard symbols. A sign language feedback framework may process video data representing conversational sign language symbol sequences to extract kinematic keypoints to search a sign language dictionary representing an established vocabulary of sign language symbols. Based on selecting an intended sign language symbol from the dictionary and computing deviations between the kinematic keypoint pattern of the intended sign language symbol from the dictionary and the extracted kinematic keypoint symbol, the sign language feedback framework may provide real-time kinematic feedback to the signer describing how to adjust their signing.
A writability monitoring circuit in a static RAM (SRAM) of an integrated circuit (IC) is disclosed. In one embodiment, the writability monitoring circuit comprises a register, an array of multi-port test bitcell configured as a ring oscillator, and control circuitry. The control circuitry is configured to apply a voltage to the array of multi-port test bitcells and determine if the ring oscillator is oscillating. In one embodiment, the control circuitry is configured to adjust the voltage applied to the array of multi-port test bitcells when the ring oscillator is determined to be oscillating and is configured to determine a correlating minimum write voltage to successfully write to the array of multi-port test bitcells corresponding to a writability of the array of multi-port test bitcells when the ring oscillator is determined to not be oscillating and then send the correlated minimum write voltage to the register.
Apparatus, systems, and techniques to modify neural network hyperparameterts of one or more neural networks. In at least one embodiment, based on an amount of information to be inferenced by one or more neural networks, one or more neural network hyperparameters of the one or more neural networks may be modified, causing the one or more neural networks to optimize inferring the amount of information.
A circuit includes clock detection circuitry to detect, at transitions of a clock signal transmitted across an alternating current (AC)-coupled unidirectional link, a first maximum value and a first minimum value of common mode-based amplitudes of the clock signal. A data receiver (RX) circuit is coupled to the clock detection circuitry and configured to receive a data signal from a data channel of the link. The RX circuit imposes a second maximum value and a second minimum value on bit transitions of the data signal based on, respectively, the first maximum value and the first minimum value obtained from the clock signal. The RX circuit also samples the data signal using the clock signal.
Systems and methods disclosed herein can be used to determine an architecture of a data center based on one or more criteria. A system generates a digital representation of a data center, simulates how the digitally represented data center would perform under a certain workflow, and determines which of the simulated data centers satisfies the one or more criteria.
H04L 41/0823 - Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
H04L 41/16 - Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
75.
GENERATIVE ARTIFICIAL INTELLIGENCE (AI) TO GENERATE INFORMATION RELATED TO A COMPLEX SYSTEM
In various examples, system, methods, and techniques for implementing a generative artificial intelligence (AI) to generate information related to a system (e.g., a manufacturing facility). In some embodiments, the generative AI generates a potential answer until the potential answer is responsive to prompt(s) provided by a requestor. In at least one embodiment, neural network(s) are used to generate output based at least on the prompt(s). The output may include an answer, a request for information, and/or a request to generate a tool. In at least one embodiment, the neural network(s) are used to generate the tool based at least on the request to generate the tool if the output includes the request to generate the tool. A potential answer includes the answer if the output included the answer, and/or the results obtained by using the tool if the tool was generated and used to obtain the results.
Apparatuses, systems, and techniques to perform inferencing using one or more neural networks with improved noise immunity are disclosed. In at least one embodiment, one or more neural networks may generate inferences, where these inferences may include measurably consistent information from the same input data irrespective of changes in environmental conditions.
Approaches presented herein reduce the presence of temporary artifacts such, as flickering, when using spatial hashing with simulation. Flickering can be avoided, at least in part, by ensuring that the same simulation points are utilized in separate executions of an algorithm, even where the execution order can vary. This can involve selecting a single representative point for each hash cell of a spatial hash map, where simulation for that hash cell will be performed for that representative point location, regardless of execution order. Both a location index and a selector value are stored for this hash map entry in a single value, where lower bits store the location index and higher bits represent the selector value. Storing the selector value in the higher-weight bits ensures an atomic maximum operation will primarily consider the selector value, and resort to the location index only in the event of equal selector values.
Diffusion and autoregressive models are representatives of likelihood-based generative models which aim to learn an underlying data distribution by maximizing the likelihood of the observed data under a parameterized probabilistic model. These likelihood-based generative models are particularly effective at performing visual tasks such as image and video synthesis. However, the maximum likelihood estimation (MILE) objective entails inherent limitations, such as forcing the learned density to spread out excessively, potentially leading to blurry samples. The present disclosure improves likelihood-based generative models by contrasting to training data, which can push the performance of likelihood-based generative models beyond the limits of MLE.
In various examples, a corrective operation may be performed based at least in part on detecting that at least one circuit is operating asynchronously with respect to a reference clock. An indication that at least one circuit operating asynchronously was detected may be generated. Upon detecting a circuit operating asynchronously, a corrective operation may be performed such that a component that receives data generated using the at least one circuit continues operating in view of the indication.
Apparatuses, systems, and techniques to identify a clock frequency at which one or more processors are to operate. In at least one embodiment, a processor performs an application programming interface (API) to cause one or more statistics corresponding to one or more activity levels of one or more processors to be indicated to one or more users.
Approaches presented herein may be used to render one or more fibers based on a total reflectance for the fiber within a scene. The total reflectance may be based on interactions between the one or more fibers and a light source. One or more sets of scattering functions may approximate different interactions between the one or more fibers and the light source to generate components of the total reflectance based on the individual interactions. The total reflectance may be used as part of one or more rendering pipelines to depict the one or more fibers within the scene.
A compute node performing a distributed light transport simulation operation on a scene may select another compute node(s) for forwarding of a ray based on determining graphical data assigned to the other compute node(s) has already been intersection-tested using the ray. Thus, the compute node can avoid forwarding the ray when the graphical data has already been processed using the ray, while providing flexibility in the partition strategy used to partition the scene amongst the compute nodes. The compute node may receive and/or determine traversal information indicating compute nodes that have already intersection-tested the ray and/or have not yet intersection-tested the ray. The traversal information may include a list of compute nodes that have or have not yet intersection-tested the ray. In some examples, the compute node replays the traversal logic used by the compute nodes to generate one or more portions of the list.
Apparatuses, systems, and techniques to execute CUDA programs. In at least one embodiment, an application programming interface is performed to indicate one or more limitations of one or more attributes of one or more groups of blocks of one or more threads.
Disclosed are systems and techniques that may generate prompts for language models. The techniques include training a hierarchical virtual token generator (VTG) model to generate a large language model (LLM) input prompt, wherein training the VTG model includes training, based at least on a first dataset, a first VTG to output a first virtual token embedding, and training, based at least on a second dataset different from the first dataset, a second VTG to output a second virtual token embedding, wherein, during the training of the hierarchical VTG model, the LLM input prompt comprises the first virtual token embedding and the second virtual token embedding.
In various examples, feature identification using language models for autonomous and semi-autonomous systems and applications is described herein. Systems and methods described herein may use a language model(s) to determine information associated with features, such as surface markings, within an environment. For example, sensor data may be used to generate one or more images or other sensor data representations corresponding to an environment. The image(s) may then be processed to generate input data (e.g., input tokens) that is applied to the language model(s). Based at least on processing the input data, the language model(s) may be trained to output data (e.g., output tokens) representing information associated with one or more features. Additionally, the output data may be used to determine the information associated the feature(s) within the environment, where the information may then be used to update a map and/or navigate one or more machines within the environment.
Apparatuses, systems, and techniques to process regular expressions. In at least one embodiment, a regular expression that includes at least one zero-length assert is refactored and processed as an equivalent plurality of regular expressions that do not contain zero-length asserts.
Apparatuses, systems, and techniques to translate information of neural network portions to perform neural network portions. In at least one embodiment, one or more translation neural networks are used to translate information used between two or more portions of two or more neural networks to perform inferencing based on the translated information.
A link training process includes training a transmitter equalizer; receiving a first block of data comprising a plurality of pages of data each associated with tap settings of the transmitter equalizer; sending a first acknowledgment message indicating correct receipt of the first block of data; and, after sending the first acknowledgment message, receiving a second block of data comprising one or more pages of data each associated with the tap settings of the transmitter equalizer.
Approaches presented herein may be used to render one or more fibers based on a total reflectance for the fiber within a scene. The total reflectance may be based on interactions between the one or more fibers and a light source. One or more sets of scattering functions may approximate different interactions between the one or more fibers and the light source to generate components of the total reflectance based on the individual interactions. The total reflectance may be used as part of one or more rendering pipelines to depict the one or more fibers within the scene.
A method includes inverting a polarity of a wired connection that couples a first peer device to a second peer device and reverting the polarity of the wired connection. Based on the inverting and the reverting of the wired connection, determining that the first peer device and the second peer device share a capability.
A method includes initiating a link training process to train a transmitter equalizer, receiving a first block of data includes a plurality of pages of data each associated with tap settings of the transmitter equalizer, sending a first acknowledgement message that indicates correct receipt of the first block of data, and after sending the first acknowledgement message, receiving a second block of data includes one or more pages of data each associated with the tap settings of the transmitter equalizer.
Disclosed are apparatuses, systems, and techniques for dynamic gaze pattern analysis for advanced operator distraction detection. The techniques include receiving two or more image frames of an operator (e.g., driver) of a machine (such as a vehicle), and determining, based on the two or more image frames, one or more gaze features of the operator. The techniques further include providing, as input to an artificial intelligence (AI) model, the one or more gaze features of the operator, wherein the AI model is trained to provide an indication of an abnormal gaze pattern of the operator. The techniques further include receiving, as output from the AI model, the indication of the abnormal gaze pattern, and determining, based on the indication of the abnormal gaze pattern, whether the operator is distracted.
The disclosure provides a hardware implemented solution for warp perspective. The disclosed warp perspective accelerator (WPA) provides improved latency and energy efficiency compared to existing software solutions. In addition to the WPA, a method of warping an image using warping hardware is disclosed. In one example, the method includes: (1) receiving an instruction to warp a source image from a computer vision application or a controller, (2) determining a source coordinate of the source image based on a destination coordinate of a destination image and a transformation matrix, (3) computing a destination value for a pixel of the destination image at the destination coordinate based on source values of one or more pixels of the source image associated with the source coordinate, and (4) sending a memory write request for the destination value to a memory, wherein at least the determining and the computing are performed by the warping hardware.
In various examples, a computing system can receive a caption regarding an image. The computing system can generate, responsive to detecting that the caption includes at least a threshold amount of text, based at least on applying text recognition to the image, text data represented in the image. The computing system can cause at least one language model to generate an annotation for the image based at least on the caption, the text data, an example query for the language model to process, and an example response corresponding to the example query.
G06V 20/70 - Labelling scene content, e.g. deriving syntactic or semantic representations
G06V 10/74 - Image or video pattern matchingProximity measures in feature spaces
G06V 10/762 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
In various examples, a computing system can generate a first edge representing an action of a subject in a three-dimensional (3D) scene with respect to an object in the 3D scene. The computing system can generate, based at least on the action and one or more bounding boxes of the subject in one or more views of the 3D scene, a caption representing the subject. The computing system can generate, based at least on 3D coordinates for the 3D scene, a second edge representing a spatial relationship between the subject and the object. The computing system can update a 3D graph of the 3D scene to include a first node that corresponds to the subject and comprises the caption and a second node that corresponds to the object, the 3D graph mapping the first edge and the second edge between the first node and the second node.
One embodiment of a method for storing and accessing data includes receiving a request that specifies a first memory address and a memory bank cluster size, and storing data in or reading data from a memory bank included in a memory based on the first memory address and the memory bank cluster size.
Apparatuses, systems, and techniques to identify information to evict from a Key-Value (KV) cache. In at least one embodiment, information stored within one or more large language model (LLM) KV caches may be identified to cause an indication to generated of information stored within the one or more LLM KV caches that can be evicted without causing other information stored within the one or more LLM KV caches to be restored to the one or more LLM KV caches.
Approaches are described for automatically determining the optimal location to inpaint objects into an image to eliminate the need for manual mask creation. Segmentation models are used to analyze and classify regions within an input image and identify suitable locations for object placement based on user prompts. Multiple candidate objects may be generated using a text-to-image model, and these objects are adjusted to fit the related regions based on user prompt. The adjustment may include adjusting object parameters including size, aspect ratio, and other contextual factors for natural integration into the scene. Post-processing steps, such as adjustments based on the image's color profile, are further applied to enhance the visual consistency and quality of the final result. Multiple final images may be generated, each presenting variations in object placement and positioning, from which the user can select one or more preferred options.
Systems and methods are directed a leak detection system that incorporates one or more chemical sensors to detect small quantities of leaks. The leak detection system may localized chemical sensors to detect leakage within a specific sub-volume of an enclosure. The one or more chemical sensors may be configured to detect an additive incorporated into cooling fluid.
G01M 3/20 - Investigating fluid tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point using special tracer materials, e.g. dye, fluorescent material, radioactive material
H05K 7/20 - Modifications to facilitate cooling, ventilating, or heating
