元の名称記録: Amazon Technologies, Inc.

Technologies directed to providing a wireless chipset with integrated radar are described. A wireless device receives a first radar frame having leakage between a transmit antenna and a receive antenna. The wireless device receives a second radar frame having leakage between the transmit and receive antennas. The second radar frame is aligned to the first radar frame based on their respective leakages. A presence of a target is detected using the aligned second radar frame.

Systems and methods are described for automating the performance of a task on behalf of a user that the user would otherwise perform manually or semi-manually. In order to automate the task, a user can provide a task automation system with a description of the steps the user would implement in order to perform the task manually. The description of the steps can be provided to the task automation system as multi-modal input. The task automation system may convert the multi-modal input into tokens and input the tokens to an AI model trained to generate a workflow description. The task automation system may later generate instructions using an AI model based on the workflow description and annotations of the application used to perform the task. The task automation system may then implement the instructions generated by the AI model to perform the task automatically on behalf of the user.

Disclosed are systems and methods that offload the work traditionally performed by a thick software client operating on each container instance of a cloud provider network with a thin and generalized agent that can be instructed in a piece-wise manner to perform operations as instructed by a workload manager executing on a control plane that is separate from the container instance. The workload manager, executing on the control plane of a cloud provider network, may precompute a set of operations and order of execution of those operations in the control plane and present those operations in a controlled manner to the agent that executes each operation as instructed. The agent executing on the data plane, rather than polling an orchestrator for work and then establishing the runtime environment based on a received Application Specification, awaits an operation or task that is pushed to the agent from the control plane.

Scalable recovery of a database is performed by a distributed data store. Storage nodes can monitor connection status of database access applications that have attached respective portions of a table stored at the storage nodes to provide access to the table. A connection failure of a database access application can be detected. Transactions associated with the failed database access application can be identified and determination of transactions that are not committed made. The storage nodes can remove versions of records of the table associated with the transaction as part of recovering the respective portions of the table from the connection failure of the database access application.

Scalable garbage collections is performed by a distributed storage system. Garbage collection events are detected by a distributed storage system for different portions of a table. Garbage collection is performed for individual ones of the different portions of the table responsive to detecting the garbage collection events, including identifying one or more versions of a record to reclaim from the different portions of the table based on transaction status information and reclaiming the one or more versions of the record.

Systems and methods are provided for receiving a vector to be inserted into a set of vectors, wherein the set of vectors is stored as a plurality of leaf objects of a distributed proximity-based graph across a plurality of storage devices of a distributed object storage system, identifying a subset of the plurality of leaf objects based at least partly on a similarity of the vector to one or more vectors stored in the subset of the plurality of leaf objects, storing the vector in an insert buffer associated with the subset of the plurality of leaf objects, and in response to a vector query, generating query results comprising the vector stored in the insert buffer, and at least one of the one or more vectors stored in the subset of the plurality of leaf objects.

A system may receive a request to generate an object. A system may generate, based on the request, a first portion of a prompt associated with a first layer of a hierarchy and a second portion of the prompt associated with a second layer of the hierarchy, where the second layer has a lower priority than the first layer. A system may generate the object based in part on providing the portions of the prompt as input to a machine learning model, where the object is formatted according to an object format and is internally consistent. To generate the object the machine learning model does not violate instructions associated with a layer of the hierarchy based on instructions associated with a layer of the hierarchy having a relatively lower priority.

Record-level transaction conflicts may be detected at storage nodes of a database back-end storage service. When a prepare statement is received, a storage node can determine whether another prepare statement acting on at least one of the same records has succeeded. If so, then the storage node can indicate to the database access application that there is a conflict, allowing the database access application to wait to try again or abort the transaction.

A write to a data set may be replicated across regions using independent encryption. Replicated writes are encrypted using payload keys specific to regions. Payload keys can be encrypted and be shared between regions using respective hierarchies of keys for each region, including shared public keys of public-private key pairs.

Rowblock modifications may be performed at record-aware distributed storage systems. Record-aware distributed storage systems may implement a storage service engine that can evaluate access requests to a table and make rowblock modification determinations based on the evaluation. The storage service engine can instruct storage nodes of the record-aware distributed storage system to perform the rowblock modifications.

Requests for records are performed at a record-aware distributed storage system. A request at a storage service engine for records received from a database access application has a time value identified corresponding to a state of a table and storage nodes that store the records. Requests are sent to the storage nodes to obtain the records. The storage nodes may identify rowblocks with respective record identifier ranges and time value ranges corresponding to the records and the time value for the access request. A result may be returned to the database access application based on the records received from the storage nodes.

Techniques for predicting an action(s) to perform for a user and, optionally, delivering proactive experiences are described. A system receives data usable to determine a predicted action of a user and invokes a generative model to process the data and determine the predicted action. The system may thereafter determine a system-performable action corresponding to the predicted action and determine a task(s) for executing the system-performable action. The system may also invoke the or another generative model to determine a trigger event(s) for triggering performance of the task(s). The system may receive an event indicating the trigger event(s) has occurred and, based thereon, perform the task(s). Alternatively, a generative model may determine proactive content is to be output during a dialog with the user and, based thereon, the system may perform the task(s).

Techniques for speech-to-text are described. Examples of a speech-to-text services are described. In some examples, the service performs speech-to-text according to the request using the speech-to-text service to generate the transcript from the audio stream by: determining a compute instance to send the request to based, at least in part, on availability information maintained in a distributed routing cache for a plurality of compute instances and types of speech-to-text processing indicated by the request, sending the request to the determined compute instance, and processing the request using the determined compute instance to generate the transcript.

A system may be configured for cooperation between language model agents. An agent may be, for example, a computer system, or a software component executing on a computer system, that can accept text and/or natural language inputs, draw upon an LM to process the inputs and perform a function, and respond via text and/or natural language outputs. An agent may act as a mediator to interact with a user, identify a task requested by the user, and delegate one or more subtasks to another agent or other resource. An agent may act as a delegate to handle tasks or subtasks delegated by a mediator. Agents may communicate with each other using a combination of structured and unstructured language; for example, one or more parameters and a natural language message.

A system may be configured for cooperation between language model agents. An agent may be, for example, a computer system, or a software component executing on a computer system, that can accept text and/or natural language inputs, draw upon an LM to process the inputs and perform a function, and respond via text and/or natural language outputs. An agent may act as a mediator to interact with a user, identify a task requested by the user, and delegate one or more subtasks to another agent or other resource. An agent may act as a delegate to handle tasks or subtasks delegated by a mediator. Agents may communicate with each other using a combination of structured and unstructured language; for example, one or more parameters and a natural language message.

The present disclosure relates to methods, apparatus, systems, and non-transitory computer-readable storage media for training and using a multi-scale machine learning model for the enhancement of compressed video. According to some examples, a computer-implemented method includes receiving a video at a content delivery service; performing an encode on a frame of the video by the content delivery service that converts the frame from a pixel domain to a transform domain and back to the pixel domain to generate first pixel values and a first residual for a block of the frame at a first resolution; generating a first set of features, by a machine learning model of the content delivery service, for an input, at a first resolution, of the first pixel values and the first residual of the block; generating a second set of features, by the machine learning model of the content delivery service, for an input, at a second lower resolution, of second pixel values and a second residual of the block; upsampling the second set of features to the first resolution to generate an upsampled second set of features; generating a modified version of the frame based on the first set of features and the upsampled second set of features; and transmitting the modified version of the frame to a frame buffer or from the content delivery service to a viewer device.

Techniques are generally described for Wi-Fi uplink and Bluetooth downlink optimization. An example method includes initiating a first counter corresponding to a first time period, executing first operation using a first radio during the first time period, and initiating a second counter corresponding to a second time period allotted for the first operation within the first time period. The example method also includes determining a second operation associated with a second radio, and in response to the determination, stopping execution of the first operation, pausing the second counter at a first counter value, and executing the second operation using the second radio during the first time period.