Abstract: A computer-implemented method may include receiving, by a tensor controller, a target loop count. The computer-implemented method may also include performing, by the tensor controller, an early exit from a static loop nest based at least in part on the target loop count. Various other devices, systems, and methods are also disclosed.

Abstract: Systems and methods for tensor cache compression and/or decompression are disclosed. An example method includes receiving a tensor or dynamically generated data. The example method includes compressing the tensor (or the dynamically generated data) by applying a compression scheme to values of the tensor (or dynamically generated data). The example method also includes storing a compressed tensor into a tensor cache (or compressed dynamically generated data into a respective cache). The example method includes reading the compressed tensor from the tensor cache (or the compressed dynamically generated data from the respective cache), and decompressing the compressed tensor (or compressed dynamically generated data) by applying a decompression scheme to values of the compressed tensor (or compressed dynamically generated data). The example method further includes forwarding a decompressed tensor (or decompressed dynamically generated data) to a compute unit.

Abstract: A robot control apparatus can include a memory that stores computer-executable instructions and at least one processor that executes the instructions by accessing the memory. The at least one processor can apply event information about activity of a user, which can be identified from user activity data perceived from a robot, and context information about time and space in which the activity occurs, to a knowledge graph formed by a relation between an event instance regarding the event information and a context instance regarding the context information, obtain user intent data regarding intent of the activity by applying the event instance among instances included in the knowledge graph to a rule creation model for creating information about the intent of the activity, and control the robot such that the robot performs a target task related to expected activity, which can follow the activity, based on the user intent data.

Abstract: The disclosed computer-implemented method may include recognizing a last instruction of a layer from a subset of a plurality of layers of a first machine learning model during its execution. The method may also include identifying a request for executing a second machine learning model and performing a context switch to the second machine learning model after executing the last instruction of the layer. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: In one embodiment, a method includes receiving a token indicating a data chunk becomes available in a first circular buffer of a pre-determined size from a direct memory access component, determining that a computation is to be performed with data including the data chunk based on the token, and generating one or more addresses corresponding to one or more data chunks within the first circular buffer that are to be retrieved for the computation, where a generated address is subtracted by the pre-determined size of the first circular buffer when the generated address is greater than a pre-determined maximum associated with the first circular buffer, and where the generated address is added by the pre-determined size of the first circular buffer when the generated address is less than a pre-determined minimum associated with the first circular buffer.

Abstract: In one embodiment, a method for accessing an instruction that is to be executed by a control agent within a computing system may include decompressing the instruction by replacing each of one or more zero-symbol run-length fields in the instruction with as many continuous zero symbols as its corresponding value and removing one or more non-zero-symbol run-length fields from the instruction. The method may also include determining that the instruction is spatial-delta-encoded based on a compression data header associated with the instruction, performing spatial-delta decoding on the instruction in response to the determination by orderly determining a spatial-delta-decoded value of each bit in the instruction, and causing the instruction to be sent to the control agent.

Abstract: Provided are a device and method for dialogue state tracking (DST). With the device and method for DST, an appropriate dialogue state for a user's intention is tracked by hierarchically classifying each slot in consideration of multiple features of domain-specific slots extracted from a dialogue with the user and tracking a state value of each of the hierarchically classified slots using a dual dynamic graph neural network.

Abstract: In one embodiment, a method includes, determining that a bmm operation between a first activation tensor and a second activation tensor needs to be performed, collecting the second activation tensor in two blocks from activation buffers of N tensor processor units, splitting each of the two blocks of the second activation tensor into an MSB tile and an LSB tile, loading the second activation tensor to weight buffers of the N tensor processor units by filling a first entry of each weight buffer of each of the N tensor processor units with contents of the MSB tiles of the two blocks and filling a second entry of the weight buffer with contents of the LSB tiles of the two blocks, and generating a bmm result using the first activation tensor distributed in the activation buffers and the second activation tensor in the weight buffers.

Abstract: In one embodiment, a method for retrieving a compressed data chunk from a source memory to a data buffer using a direct-memory access includes generating a source address indicating a location in the source memory at which a metadata corresponding to a compressed data chunk is stored, reading the metadata from the source address, where the metadata includes a data address, a size and compression options associated with the compressed data chunk, reading the compressed data chunk from the source memory based on the data address and the size within the metadata, decompressing the compressed data chunk based on the compression options within the metadata, and storing the decompressed data chunk into the data buffer.

Abstract: In one embodiment, a method for accessing an instruction that is to be executed by a control agent within a computing system, decompressing the instruction by replacing each of one or more zero-symbol run-length fields in the instruction with as many continuous zero symbols as its corresponding value and removing one or more non-zero-symbol run-length fields from the instruction, determining that the instruction is spatial-delta-encoded based on a compression data header associated with the instruction, performing spatial-delta decoding on the instruction in response to the determination by orderly determining a spatial-delta-decoded value of each bit in the instruction, and causing the instruction to be sent to the control agent.

Abstract: In one embodiment, a method includes receiving a token indicating a data chunk becomes available in a first circular buffer of a pre-determined size from a direct memory access component, determining that a computation is to be performed with data including the data chunk based on the token, and generating one or more addresses corresponding to one or more data chunks within the first circular buffer that are to be retrieved for the computation, where a generated address is subtracted by the pre-determined size of the first circular buffer when the generated address is greater than a pre-determined maximum associated with the first circular buffer, and where the generated address is added by the pre-determined size of the first circular buffer when the generated address is less than a pre-determined minimum associated with the first circular buffer.

Abstract: In one embodiment, a method includes, at each iteration i among N iterations of a first loop, reading first data corresponding to row i of a first tensor from a first source memory, reading second data from column i of the transpose buffer, writing the first data to column i of the transpose buffer, and causing the second data to be written to row i of a second tensor at a first destination memory and, at each iteration j among N iterations of a second loop, reading third data corresponding to row j of a third tensor from a second source memory, reading fourth data from row j of the transpose buffer, writing the third data to row j of the transpose buffer, and causing the fourth data to be written to row j of a fourth tensor at a second destination memory.

Abstract: In one embodiment, a method includes, determining that a bmm operation between a first activation tensor and a second activation tensor needs to be performed, collecting the second activation tensor in two blocks from activation buffers of N tensor processor units, splitting each of the two blocks of the second activation tensor into an MSB tile and an LSB tile, loading the second activation tensor to weight buffers of the N tensor processor units by filling a first entry of each weight buffer of each of the N tensor processor units with contents of the MSB tiles of the two blocks and filling a second entry of the weight buffer with contents of the LSB tiles of the two blocks, and generating a bmm result using the first activation tensor distributed in the activation buffers and the second activation tensor in the weight buffers.

Abstract: In one embodiment, a method for iteratively transferring a plurality of non-contiguous blocks of data from a source memory to a destination memory through n-dimensional loops without being re-programmed by a direct memory access within a machine-learning accelerator includes reading a first block of data from a first address of the source memory, processing the first block of data with an ingress modification function, and storing the first block of data to a second address of a data buffer, by an ingress component of the direct memory access within the machine-learning accelerator, and reading a second block of data from a third address of the data buffer, processing the second block of data with an egress modification function, and storing the second block to a fourth address of the destination memory, by an egress component of the direct memory access within the machine-learning accelerator.

Abstract: A circuit for performing computations for a neural network comprising multiple neural network (NN) layers. The circuit includes a processing device that provides programming data for performing the computations and a core in data communication with the processing device to receive the programming data. The core includes activation memory that stores inputs for a layer and parameter memory that stores parameters for a first NN layer. The core also includes a rotation unit that rotates accessing the inputs from the activation memory based on the programming data and a computation unit that receives a respective input and a parameter for the first NN layer and generates an output of the first NN layer using the input and the parameter. The core also includes a crossbar unit that causes the output to be stored, in the activation memory, in accordance with a bank assignment pattern.

Abstract: Computers and methods for performing mathematical functions are disclosed. An embodiment of a computer includes an operations level and a driver level. The operations level performs mathematical operations. The driver level includes a first lookup table and a second lookup table, wherein the first lookup table includes first data for calculating at least one mathematical function using a first level of accuracy. The second lookup table includes second data for calculating the at least one mathematical function using a second level of accuracy, wherein the first level of accuracy is greater than the second level of accuracy. A driver executes either the first data or the second data depending on a selected level of accuracy.

Abstract: Disclosed are an apparatus and a method of searching for information by a mobile device. The present invention provides simplified ontology to be applicable to a mobile environment having a limited resource, and provides ontology capable of providing a combined search environment by combining DBs used by various applications within a mobile device, respectively. Further, the present invention provides a semantic search engine providing a function of searching a local database within a mobile device, and expanding a search to a web as necessary and performing the search.

Abstract: A wearable device is provided for authentication that includes a memory element and processing circuitry coupled to the memory element. The memory element configured to store a plurality of user profiles. The processing circuitry is configured to identify a pairing between the wearable device and a device. The processing circuitry is configured to identify a user of the wearable device. The processing circuitry also is configured to determine if the identified user matches a profile of the plurality of user profiles. The processing circuitry is also configured to responsive to the identified user matching the profile, determine if the profile provides authorization to access the device. The processing circuitry is also configured to responsive to the profile providing authorization to the device, send a message to the device authorizing access to the device.

Abstract: A thinner composition includes propyleneglycol alkylether acetate, cycloketone, and methyl 2-hydroxy isobutyrate. The thinner composition has excellent EBR, RRC, and rework properties, as well as effects of improving photoresist application uniformity, and in particular, exhibiting excellent solubility to photoresist having a high polarity, so as to be applicable to various photoresists.

Abstract: A thinner composition includes propyleneglycol C1-C10 alkylether, C1-C10 alkyl C1-C10 alkoxy propionate, and C1-C10 alkyl lactate, thereby is it possible to improve applicability of a photoresist while remarkably reducing an amount of the used photoresist, as well as having excellent solubility and EBR property to various photoresists, BARCs and underlayers.