Abstract: Described are examples for detecting anomalies in files. Each of multiple files can be processed to generate corresponding flat files. Values can be extracted from multiple lines of each of the flat files into a parameter vector. Patterns can be generated from the multiple lines based on the values extracted. The parameter vector and the patterns can be provided as input to a machine learning (ML) model to obtain a set of rules for the files based on conditional probabilities. The set of rules can be applied to a set of one or more files to detect anomalies in values in the set of one or more files.

Abstract: Embodiments of this application provide a DC/DC conversion circuit, a power unit, a charging pile, and a charge-discharge heating method. The circuit includes: a first rectifier module, where an input end of the first rectifier module is connected to a power grid through an AC/DC conversion circuit; a transformer module, where an input end of the transformer module is connected to an output end of the first rectifier module; an energy storage module; and a second rectifier module, where an input end of the second rectifier module is configured to connect to an output end of the transformer module or the energy storage module, and an output end of the second rectifier module is configured to connect to a battery pack of an electric vehicle when the charging pile is charging the electric vehicle.

Abstract: Methods and apparatuses for improving the performance and energy efficiency of Radio Access Networks (RANs) are described. Various power control schemes may dynamically adjust RAN power consumption based on fluctuations in network traffic, throughput, latency, queue sizes, and/or packet error rates with the goal of increasing energy efficiency while maintaining quality of service metrics. The power control schemes may be implemented using a PRB controller for dynamically allocating physical resource blocks (PRBs) to user devices and a CPU controller for assigning CPU power profiles based on PRB allocations for the user devices. The PRB controller and CPU controller may periodically acquire real-time telemetry data and wireless network performance information and then adjust the number of PRBs for user devices and adjust the CPU power profiles for executing RAN functions based on the telemetry data and wireless network performance information.

Abstract: Computing technology is described herein that provides an attention mechanism, implemented by a neural network, that generates attention information based on head-specific query information and shared key and value (KV) information, without computing head-specific key information and head-specific value information, and without caching the head-specific key information and the head-specific value information in memory. This manner of operation allows the computing technology to make efficient use of processing and memory resources. In some implementations, the attention mechanism is part of decoder of an encoder-decoder system, or a standalone decoder system. In some implementations, the computing technology leverages the attention information to generate synthesized text based on input text.

Abstract: A motor controller includes: a plurality of first power modules, a second power module, a first group of connecting interfaces, a second group of connecting interfaces, and a first capacitor, where the plurality of first power modules are connected in parallel to each other, and the plurality of first power modules are configured to connect to the first group of connecting interfaces; the second power module is configured to connect to the second group of connecting interfaces; and the first capacitor is connected in parallel to the second power module, where the plurality of first power modules and the second power module are configured to perform temperature regulation on an object connected to the first group of connecting interfaces and the second group of connecting interfaces.

Abstract: Systems and methods for dynamically modifying a cache associated with a neural network model of a natural language generator are described. In examples, a neural network model employs a beam search algorithm at a decoder when decoding output and generating predicted output candidates. The decoder utilizes caching techniques to improve a speed at which the neural network operations. When an amount of memory utilized by one or more caches of the neural network model is determined to exceed a threshold memory size, a layer-specific portion of a cache associated with a layer of the neural network model is identified. The identified layer-specific portion of the cache can be deleted when the amount of memory utilized by the cache of the neural network model exceeds the threshold memory size. In examples, data in the cache is deduplicated and/or deleted.

Abstract: A battery main box, a charge-discharge circuit, and a powered device. The battery main box includes two first electrode circuits and one second electrode circuit. A switch is connected between the two first electrode circuits. The two first electrode circuits are respectively configured to be connected to two powered loads and two batteries, and the second electrode circuit is configured to be respectively connected to the two powered loads and the two batteries. The battery main box in the embodiments of the present application includes the two first electrode circuits and the one second electrode circuit, and the switch is connected between the two first electrode circuits.

Abstract: A charge and discharge control method comprises: when a battery heating condition is satisfied, turning on a connecting circuit between a neutral point of a first electric motor and a neutral point of a second electric motor, wherein a first electric motor controller connected to the first electric motor and a second electric motor controller connected to the second electric motor are both connected between a positive electrode and a negative electrode of a battery; and controlling the turning-on/turning-off of bridge arms in the first electric motor controller and the second electric motor controller, and forming, in a circuit in which the battery is located, a charging loop and a discharging loop which are alternately switched.

Abstract: Embodiments of the present application provide a charging and discharging apparatus and a battery charging method, which are capable of ensuring security performance of a battery. The apparatus comprises a bi-directional DC/DC converter and a control unit, wherein the control unit is configured to: receive a first charging current transmitted by a battery management system (BMS) of a battery, control the bi-directional DC/DC converter based on the first charging current to charge the battery through an energy storage battery; receive a first discharging current transmitted by the BMS and control the bi-directional DC/DC converter based on the first discharging current to discharge a battery capacity of the battery to the energy storage battery; and receive a second charging current transmitted by the BMS and control the bi-directional DC/DC converter based on the second charging current to charge the battery through the energy storage battery.

Abstract: A charge-discharge circuit and an electric device. The charge-discharge circuit includes a power supply module, a first drive assembly, a second drive assembly, and a first switch module, where the first drive assembly and the second drive assembly are connected in parallel between a positive terminal and a negative terminal of the power supply module; and a terminal of the first switch module is connected to the first drive assembly, and another terminal of the first switch module is connected to the second drive assembly.

Abstract: A charging and discharging circuit, a battery control circuit and an electrical device. The charging and discharging circuit includes: a power supply module, a first driving assembly, a second driving assembly and a first switch; the power supply module includes a battery, the first driving assembly includes a first motor, and the second driving assembly includes a second motor; the first driving assembly and the second driving assembly are connected in parallel between a positive electrode end and a negative electrode end of the power supply module; and one end of the first switch is connected to a neutral point of the first motor, and the other end of the first switch is connected to a neutral point of the second motor.

Abstract: A digital ocular system for use with image sensors and an ophthalmic microscope includes a housing that connects to the microscope and defines a housing cavity, a first lens having a first optical axis, and a second lens having a second optical axis parallel to the first optical axis. A first micro-display is positioned within the housing cavity to present a first image from the image sensors. The first micro-display has a first center axis arranged at a predetermined angle with respect to the first optical axis. A second micro-display presents a second image from the image sensors. The second micro-display includes a second center axis arranged at the predetermined angle with respect to the second optical axis, such that the first and second micro-displays are horizontally decentered relative to the first and second optical axes, respectively.

Abstract: This application provides an energy storage system control method, a control apparatus, and an energy storage system. The energy storage system is coupled to an energy generation system and includes a plurality of energy storage units. The energy storage system control method includes: controlling, in a first stage, the energy generation system to charge a first energy storage unit in the plurality of energy storage units, where the first stage is a charging stage in a charging cycle of the first energy storage unit; and controlling, in a second stage, the first energy storage unit to discharge electricity to a second energy storage unit in the plurality of energy storage units, where the second stage is a discharging stage in the charging cycle of the first energy storage unit.

Abstract: Systems and methods are provided for training a data model based on training data. The training includes pre-training and fine-tuning the data model based on a combination of an autoregressive (AR) model and a non-autoregressive (NAR) model. Training data may be received and encoded into streams of tokens. A pre-trainer during decoding generates a continuum of data structures of the AR and NAR combined model including a main stream and a series of predicting streams. Masked tokens in predicting streams reference or attend to one or more preceding tokens in the main stream or the preceding predicting streams. A fine-tuner selects streams to generate a trained model according to a target data model. The target data model is determined based on balancing an accuracy constraint and an efficiency constraint for predicting tokens. The decoder acts as abridge between the AR and NAR models in generating a trained data model.

Abstract: The present disclosure provides a method and a device for evaluating a shipping risk caused by a sea ice disaster, and a computing device. The method includes: obtaining a historical daily-scale sea ice thickness data set of a monitored region; evaluating a risk level of sea ice disaster-inducing factors, vulnerability of a disaster-affected body and an exposure level in accordance with the historical daily-scale sea ice thickness data set; determining a disaster-bearing capability and a risk level of the disaster-affected body in the sea ice disaster in accordance with the risk level of the sea ice disaster-inducing factors, the vulnerability of the disaster-affected body and the exposure level; and evaluating shipping risks for different types of disaster-affected bodies within different return periods in accordance with the disaster-bearing capability and the risk level of the disaster-affected body in the sea ice disaster.

Abstract: A tunable eye model includes a housing having an inner chamber and at least one aperture adjacent to an anterior portion of the inner chamber. A lens is at least partially disposed in the at least one aperture, and the lens has a variable focal length. An optical target of the tunable eye model is disposed over a posterior portion of the inner chamber. The optical target is configured to facilitate evaluation of a characteristic of an ophthalmic microscope when viewed through the lens with the ophthalmic microscope.

Abstract: A motor controller includes: a plurality of first power modules, a second power module, a first group of connecting interfaces, a second group of connecting interfaces, and a first capacitor, where the plurality of first power modules are connected in parallel to each other, and the plurality of first power modules are configured to connect to the first group of connecting interfaces; the second power module is configured to connect to the second group of connecting interfaces; and the first capacitor is connected in parallel to the second power module, where the plurality of first power modules and the second power module are configured to perform temperature regulation on an object connected to the first group of connecting interfaces and the second group of connecting interfaces.

Abstract: Provided are a power battery voltage regulation circuit and system, and a control method and a control apparatus therefor. The power battery voltage regulation circuit includes a power module, a heating module, a charge/discharge interface, and a voltage regulation switch assembly, wherein the heating module comprises an energy storage element and a switch module; the power module comprises at least a first power battery and a second power battery; the power module is connected in parallel with the switch module; and an external charge/discharge device is connected in parallel with the power module through the charge/discharge interface; the voltage regulation switch assembly comprises a plurality of switches that are provided between the charge/discharge interface and the power module; and the voltage regulation switch assembly and the switch module are used for regulating, in response to a control signal, a charge/discharge voltage between the external charge/discharge device and the power module.

Abstract: An accelerated attention mechanism with parallel operations can improve machine learning technology by enabling execution of certain matrix multiplication operations in parallel with element-wise operations, leading to an increase in speed without quality loss. To compute attention values in a machine learning model, the mechanism can receive a query vector, key vector, and value vector and split each of these vectors into blocks. For a given query block, the mechanism can determine attention values by performing element-wise operations to update the attention values for the given query block based at least in part on previously computed attention scores for the given query block and a given key block. Concurrent with performance of at least some of the element-wise operations, the mechanism can perform a matrix multiplication operation using given query block and a next key block to determine attention scores for the given query block and next key block.

Abstract: A battery control module, a battery pack, and a battery control system are provided. The battery control module includes a first circuit, a second circuit, and a third circuit. The first circuit is connected to a positive electrode of a first battery module, the second circuit is connected to a positive electrode of a second battery module, and negative electrodes of the first battery module and the second battery module are both connected to the third circuit. The first circuit, the second circuit, and the third circuit are connected to a motor control circuit. Each circuit is a circuit with an adjustable on-off state, to switch a circuit connection relationship between the first battery module and the second battery module with the motor control circuit.